Romanian-language GCSE InitiativeMapping Communities for Impact

ABC·H The letters in the bar at the top of the page follow the eight analytical sections A–H that come after the prologue, the introduction and the note on data. Click a letter to jump to that section; the thin line beneath shows how far through the report you are.
Executive summary

There is no GCSE in Romanian. Nineteen modern languages have one, including Polish, Urdu and Arabic. The evidence in this report says the candidates for a Romanian GCSE are already in England’s schools.

  1. 01Second language of England and Wales. 471,954 people gave Romanian as their main language in the 2021 Census, up from 67,586 in 2011. Romanian rose from nineteenth to second place, the largest increase of any language.
  2. 0298,210 pupils, three times as many as in 2016. In January 2026, England’s schools recorded 98,210 pupils with Romanian as their first language, up from 32,143 in 2016. Every compulsory year group, from Reception to Year 11, now has more than 5,000.
  3. 03The first GCSE cohort is already in school. 5,316 Romanian pupils are in Year 11 now, above the c. 5,000-candidate benchmark. Polish pupils enter the Polish GCSE at about eight per ten Year 11 pupils. At that rate, Romanian pupils would make about 4,200 entries a year, and the year groups behind it are larger still.
  4. 04Two routes give the same scale. The entry rates of comparable community languages, applied to the census speaker base, imply 5,700 to 12,200 entries at maturity. Counting enrolled pupils gives the same order of magnitude.
  5. 05The census sees half the children. Pupils with Romanian as first language outnumber the census count of school-age speakers (44,323) by about 2.2 to one. The difference is mostly the bilingual second generation, whom a single main-language question cannot count.
  6. 06Delivery has a map. In the Thurrock–Northamptonshire corridor, Romanian pupils are 12–17% of multilingual secondary pupils, enough for mainstream schools to teach and examine the subject. In large diverse cities the share is 2–3%, where the supplementary-school and private-candidate route used for Polish and Arabic applies. Hertfordshire holds the most Romanian pupils of any education authority (4,070).
Prologue: How many Romanians are in the United Kingdom

Since 2019, Romanian nationals have made nearly two million applications to the EU Settlement Scheme – more than any other nationality in Europe. However, applications do not equate individuals: the same person can apply more than once, most often on converting from pre-settled status to settled status, and Romanians are the scheme’s most frequent repeat applicants (532,200 by June 2026, again the highest of any nationality). Netting these out, roughly 1.5 million individual Romanians have applied, and somewhat over one million have been granted residence rights – an upper bound of everyone the United Kingdom has, at some point since 2019, recognised as resident.1

If we zoom in further on these fluid communities, not all remain. Romanian migration in the UK has been demonstrably circular. In the latest estimates from the Office for National Statistics (ONS), Romanian is the most common EU nationality both arriving in and leaving the UK, five years after free movement ended.2 The settled core may be the firmest layer of the resident population: over 686,000 Romanian nationals hold settled status – five years of proven, continuous residence, concluded by the Home Office itself. This figure exceeds every census reporting of the community: 539,000 born in Romania, 471,954 reporting Romanian as their main language (England and Wales, March 2021).

Two further populations sit outside all of these counts, but merit attention. Romanians who have become British – 8,285 citizenship grants in the year to June 2026 alone, and the second-largest EU nationality (behind Italy). And, larger still, the children born here: a child born in the UK to a settled parent is British at birth, and so never appears in the Settlement Scheme or the nationality counts. Births to Romanian-born mothers in England and Wales rose from 1,248 in 2007 to over 16,000 a year by 2019. In 2021, Romania was the single most common country of birth for mothers born outside the UK.3

One repository however, becomes most illuminating, and counts this generation directly based on the school and pupil censuses. In January 2026, England’s schools held 98,210 pupils with Romanian recorded as their first language – more than double the number of school-age Romanian speakers the 2021 census could observe. Irrespective of the adults’ demographic totals, this generation of pupils is already enrolled, counted and sitting in British classrooms as this report is being drafted.4

Assembling this patchwork of demographic reporting, an account of how many Romanians actually are in the UK, may read as follows: an upper bound above one million granted residence; a core of 686,000 permanently settled residents; a naturalised cohort in the tens of thousands and a British-born second generation in the low-to-mid hundreds of thousands, both invisible to the migration statistics; and most importantly to everything that follows, nearly 100,000 children with Romanian as first language, already in the school system. On any assembly of these components, the Romanian-origin community of the United Kingdom stands between roughly 700,000 and one million people. In the analysis that follows, we deliberately work from the more conservative of these figures – the 2021 ONS census counts as the most granular instrument available: the census resolves the community by language, age and all 331 local authorities, where education provision and language delivery decisions occur. The report then cascades through the variables the instruments allow: main language, since its 471,954 reported speakers in 2021 represent active daily use and, in family households, active transmission to the next generation (a thread covered in sections A, F and H); its distribution throughout the 331 local authorities; age (of Romanian speakers, as reported main language) through a census table commissioned for this report; the pupils’ first language, by year group, across 153 education authorities through which schooling is actually administered; the share of each authority’s multilingual school population; as well as, proficiency in English. With each variable we narrow the lens, from community, to area, to cohort (year group of pupils), to classroom.

1,998,079
applications to the EU Settlement Scheme by Romanian nationals since 2019 – more than any other nationality
Source: Home Office, Immigration system statistics, year ending June 2026, EUSS_D01. Applications, not people.
≈1,500,000
individual Romanians who have applied, once repeat applicants are netted out
Source: authors’ netting of repeat applicants (532,200 Romanian repeat applicants to 30 June 2026; Home Office, EUSS quarterly statistics, tables EUSS_RA_01–04): at most 1,465,879 people. The Home Office publishes no count of unique persons by nationality.
>1,000,000
granted residence rights – an upper bound on everyone the UK has recognised as resident since 2019
Source: Home Office, EUSS_D02, June 2026: 1,412,533 grants (686,129 settled, 726,404 pre-settled), net of repeat applicants; authors’ calculation.
686,129
hold settled status – five years of proven, continuous residence, concluded by the Home Office itself
Source: Home Office, EUSS_D02, applications concluded with settled status, nationality = Romania, to 30 June 2026.
471,954
report Romanian as their main language in England and Wales (539,000 were born in Romania)
Source: ONS, Census 2021 (21 March 2021), main language, usual residents aged 3 and over; country of birth as published (rounded).
98,210
pupils in England’s schools with Romanian recorded as their first language
Source: DfE, response to FOI-2026-0093642, state-funded schools, January 2026 school census.

On any assembly: roughly 700,000 to one million people. This report works from the smallest figures.

Citizenship grants, cumulative ⟨updating⟩ Births to Romanian-born mothers, cumulative ⟨updating⟩
Introduction & relevance

The Romanian presence in the United Kingdom is one of the most significant demographic developments recorded between the last two UK censuses yet remains underexplored. A community numbering some 80,000 Romanian-born residents in 2011 grew to 539,000 by March 2021 the largest increase of any country of birth over the decade – thus making Romania the fourth most common non-UK country of birth in England and Wales, while 477,000 residents declared a solely Romanian national identity, the second most common non-UK identity after Polish. This upward trajectory closely followed the chronology of policies: Romania’s EU accession in 2007 was followed by seven years of transitional labour-market restrictions, so that more pronounced settlement in the UK began after their lifting in January 2014 (a full decade behind the post-2004 Polish migration wave, with which it is often but erroneously conflated).

The linguistic consequence is captured in the 2021 Census: Romanian rose from the nineteenth to the second most common language in England and Wales (67,586 to roughly 471,954 speakers, a statistical estimate that may yet be conservative depending on the source). Because this is a relatively young, predominantly first-generation community in which intergenerational language transmission remains high, its school-age population carries a linguistic competence of a particular kind (i.e.: current, more often already acquired in the home and deployed in daily use), which the qualification system has yet to transpose into formal recognition. A decade of growth has produced not a transient workforce but a strategic constituency for the UK, economically active, regionally embedded and raising its children within the English school system – the latter also qualifying as the most palpable indicator of permanence.

Thus, it is at this intersection of demography and education that the case for a Romanian-language GCSE arises. The argument advanced in this visual and interactive report rests on convergence. The first is demographic: a settled community of national significance, whose school-age population is clustered in identifiable local-authority concentrations where teaching provision and examination infrastructure would be viable. The second condition is educational. Our independent community surveys show higher degrees of fluency acquired from domestic milieus, maintaining heritage connections with the language. For nineteen modern languages, including Polish, Urdu and Arabic, formal recognition is already available; a pupil fluent in the language of the home can sit the examination and have that fluency certified. For Romanian, no such equivalent exists yet, and the proficiency of 98,210 pupils goes officially unrecorded – an asymmetry of recognition between comparable communities. The third converging condition is institutional. Modern-languages offer has never been static, instead being periodically renewed to reflect the linguistic composition of the country it serves. Most recently, this was achieved through the protection and redevelopment of twelve community language qualifications in 2016 and the introduction of a GCSE in British Sign language.5

For nineteen modern languages, formal recognition is already available. For Romanian, no such equivalent exists yet.

Against the backdrop of this record, extending the offer to the second most spoken main language in England and Wales is consistent with, rather than a departure from, the system’s own history. Heritage languages have not been treated thus far as a concessionary annex to the national offer but a constituent part of it, a position that is reflected across the qualifications framework itself, in which community-language GCSEs carry full weight in school performance measures, including the languages pillar of the English Baccalaureate, on equal terms with French, German and Spanish.6

The data

The analysis rests on bespoke extracts from the 2021 Census of England and Wales, collected through the Office of National Statistics (ONS) and its custom dataset facility against the detailed main-language classification (TS024) at a lower-tier local-authority level, covering 331 authorities. The collections reconcile to within two persons of the published national figure (471,952 against 471,954), the residual reflecting the cell-key perturbation applied to all census outputs under statistical disclosure control (figures below are therefore reported as enumerated but should be read as accurate to within rounding). School-level evidence is drawn from the DfE’s (Department for Education) January school censuses as published in Schools, pupils and their characteristics (Explore Education Statistics), 2015/16 – 2025/26, at local-authority level by phase. Furthermore, qualification-entry figures are drawn from Joint Council for Qualifications tables for summer 2024 (UK, all awarding bodies) and Ofqual provisional entries for summer 2025 (England).

Albeit not exhaustive, two measurement properties shape our interpretation, pending the inclusion of additional data for granularity, that is not publicly available yet. The census records main language, a single response, and therefore excludes bilingual residents – most consequentially UK-born children of Romanian-speaking households who report English; all population figures below are accordingly lower bounds on the heritage-speaker cohort. The two data families also measure different moments in time – a March 2021 population snapshot against annual school censuses and summer examination cohorts. Consequently, cross-source ratios are presented as standardised indices and illustrations rather than forecasts.

Certain limitations to full visibility also exist. The statistical systems protect confidentiality by different disclosure mechanisms, with various implications. Census outputs are perturbed, which means that controlled noise is added to every cell, so all values are published but small ones are deliberately imprecise, and custom extractions (such as ours) do not sum exactly to published totals (hence the two-person residual mentioned above). No inference in this report rests on census cells smaller than ten. Correspondingly, school-census outputs are instead suppressed – small counts are withheld outright and the remainder rounded.7

Another structural difference concerns geography. Census extracts are produced at lower-tier local-authority level (totalling 331 authorities); meanwhile the Department’s school statistics are published at education-authority level (153 upper-tier authorities): in shire counties, several census districts correspond to one education authority. Every priority authority in this analysis is a unitary authority or London borough, identical in both geographies, so the cross-source comparisons in the following sections can be reconciled or joined exactly.8 Our data dictionary is rendered in the Table below. To this repository, the present iteration already embeds the datasets obtained through freedom of information (FOI) and commissioned-table requests – the Department for Education’s (DfE) named-language pupil counts (FOI-2026-0093642) and the Office for National Statistics (ONS) local-authority age table (CT21_0503), both present in the following sections. Further research iterations will add the community survey, valuable for a more granular assessment still.

ONS Census 2021 DfE school census JCQ & Ofqual entries Home Office EUSS detailed tables, June 2026 (applications, not persons) DfE FOI-2026-0093642 ONS CT21_0503
Table 1
Data dictionary
DatasetSource and vintageUnit and coverageRole in the analysisKnown limitations
Main language — exact national counts, 2011 & 2021ONS Census 2021 bulletin, Figure 1 data downloadUsual residents aged 3+, E&W; 93 languages, both-census ranksNational trajectory (A1); growth across the top tenMain-language definition; 2011 classification harmonised by ONS
Main language by local authorityONS TS024 via Create a Custom Dataset; bulletin Fig. 3 download331 lower-tier LAs × detailed language; counts and per centLA ranking; map (B1); concentration (B2); Polish comparison (B3)Cell-key perturbation; LA totals do not sum to published figures
Main language × age, by local authorityONS commissioned table CT21_0503, produced to this report’s specification (created 15 October 2026)Main language (67 categories) × 17 age bands × 331 lower-tier LAs; usual residents aged 3+Age structure (F1); school-age counts and first-examination cohort (F2, F3); census side of D4Cell-key perturbation; national figures are sums of LA cells (471,944 against 471,954 published)
Main language × English proficiency × ageONS CT21_0418 (England and Wales), publicly available commissioned tableDetailed language × proficiency × age bands, nationalEnglish proficiency (F)National only
Pupils by first language (aggregate)DfE, Schools, pupils and their characteristics, table-tool extract151–153 education authorities × phase × language; 2015/16–2025/26EAL trend (D1); denominator of the measured share (E1)Aggregate only (English / other than English / unclassified); upper-tier geography differs from census
Pupils with Romanian or Polish first language (named)DfE, response to FOI-2026-0093642Headcounts by NC year group (R–13) × 153 education authorities, state-funded schools, England; January 2016, 2021, 2026. Romanian = RMN, RMNM, RMNR; Polish = POLYear-group structure (D2); pupil map (D3); census comparison (D4); measured share (E1); Year 11 check (C)England only; upper-tier geography; three census points, not an annual series; supplied unrounded, no suppression
GCSE entries by languageJCQ, Other MFL entries, 2024 (UK); Ofqual provisional entries, 2025Entries per language per yearProvision anomaly (C1); entry-rate standardisation; entry trends (C2)JCQ covers UK, Ofqual England; all-age entries, not one year-group
EU Settlement Scheme grantsHome Office, Immigration system statistics, EUSS detailed tables (EUSS_D01/D02), June 2026Applications and concluded outcomes by nationality, sex and age; UK, cumulative 2018–2026Contemporary scale (A2); child population (F3)Counts applications, not persons; nationality, not language; UK-wide, not England and Wales
Local-authority boundariesUK LA geometries, consolidated by the authors to 2021 geography331 LTLA polygons matching census codes; dissolved by the authors into 153 education authorities (2026) with the ONS LTLA–UTLA lookupMap base layers (B1, D3)Generalised boundaries; cartographic use only
Derived tablesAuthors’ calculations from the aboveLA ranking; EAL trends; pupil-to-census ratios; measured share; retired intensity indexB2, B3, C1, D1–D4, E1Inherit all limitations above
Census cells are perturbed (no inference on cells < 10). Published school-census aggregates are suppressed and rounded; the named-language extract (FOI-2026-0093642) was supplied unrounded and contains no suppressed cells.
Analytical insights – mapping perspectives
ASection A

The national trajectory

Between 2011 and 2021 Romanian rose from the nineteenth to the second most common main language in England and Wales, from 67,586 to 471,954 speakers, registering a seven-fold increase, and the largest recorded for any language over the intercensal decade.

To place into context, the composition of the top ten languages reinforces this structural landscape. Polish, the largest community language, grew 12% from an already mature base; the South Asian languages that anchor the existing community-language qualification offer were static or mildly increasing (i.e.: Panjabi, +6.4%; Urdu, +0.4%) or in absolute decline (i.e.: Bengali, -10%; Gujarati, -11%) – the latter trend may be indicative of a generational language shift in long-settled communities.

Figure A1
The ten most common main languages other than English, England and Wales, 2011 and 2021
Usual residents aged three years and over, reporting each language as their main language, both censuses. Romanian rose from 67,586 speakers (nineteenth) to 471,954 (second) – the largest increase recorded for any language over the intercensal decade. The census records a single main language per person; figures for long-settled communities therefore may understate the full speaker populations. Source: ONS, Census 2021, data download (exact counts); percentage changes are the author’s calculations. Hover a bar for exact counts.

The table describes two cohorts of languages at different points in a settlement lifecycle of these communities. The South Asian languages entered the British examination landscape through post-war and Commonwealth migration and now belong to communities in their third and fourth generations or even upwards.9 Research literature from the Linguistic Minorities Project onward has amply documented the transformation of linguistic repertoires across generations, and among the UK-born the heritage language typically ceases to be the primary language and contracts in domains of use – sustained in the family, the community institution and the complementary school, interwoven with English in the hybrid practices of daily interactions – rather than disappearing.10

For present purposes the point is more comparative. Romanian today emerges at the opposite pole of this trajectory, accelerating after the country’s EU accession in 2007, but also post-pandemic in a cycle of circular migration and re-settlement (itself difficult to capture). Overall, growth occurred within seven years of the census date. There is one measure that should also inform our reading of such dynamics: the census records a single main language per person, so for long-settled communities – where many who speak the heritage language report English – the bars may understate the true speaker population, while the Romanian figure, drawn from a more recent economic migration wave, carries little of this undercount as yet. For now, however, aside from English, only Polish ranks above Romanian – a lead of some 140,000 held by a community two decades longer in formation.

An ONS analysis links the rise in Romanian main-language reporting directly to the parallel increases in Romanian-born residents (80,000 to 539,000, the largest increase for any country of birth) and in Romanian-only national identity (73,000 to 477,000) – that is, to first-generation migration rather than any changes in reporting behaviour.11 Administrative data corroborate this at a larger scale; by June 2026, Romanian nationals had received more than 1.4 million grants of status under the EU Settlement Scheme (the largest of any nationality) including over 686,000 in settled status, a count of proven five-year residents that exceeds the census estimates widely deployed in this report.12

Figure A2
Three measures, one migration: birth, identity and language, 2011 to 2021
Romanian-born residents, residents with a Romanian-only national identity, and Romanian main-language speakers aged three and over, England and Wales. The three measures move together, which points to first-generation settlement rather than a change in how people answered. Source: ONS, Census 2011 and 2021 (country of birth and national identity rounded as published by ONS; main language exact).

Another interesting property of this language community is youth. A population formed principally by recent labour migration should carry a school-age component proportionally larger and a higher rate of home-language transmission. The DfE’s analysis points in a similar direction, finding that 52% of pupils with Romanian first-language had first entered an English school after reception year (the signature of recently arrived families).13 Whether the age structure in fact bears the weight is tested against the census main-language-by-age cross-tabulation in the following sections.

BSection B

Geographical distribution(s)

The (speaker population) settlement geography suggests two distinguishable formations.

The first is a contiguous arc of outer London boroughs: Harrow with 18,987 Romanian-language speakers, 7.5% of the borough’s population and the highest proportion in England; Newham (17,694; 5.3%), Brent (15,570; 4.8%), Redbridge (12,828; 4.3%), Waltham Forest (10,573) and Barking and Dagenham (9,988; 4.8%). The second is a Midlands-Home Counties corridor in which Northamptonshire emerges as one of the main findings of our local-authority analysis: North and West Northamptonshire together encompass 23,330 Romanian speakers, the largest concentration in the country outside London, exceeding any borough, followed by substantial secondary clusters in Birmingham (10,573), Luton (7,455), Coventry (7,145), Doncaster (5,428), Milton Keynes (5,269) and Thurrock (4,893).

Figure B1
Romanian main-language speakers as a share of residents, by local authority, Census 2021
Show
    331 lower-tier local authorities in England and Wales (2021 boundaries); Greater London enlarged. Callouts: speakers · % of population. The alternative views outline the authorities where Romanian speakers outnumber Polish speakers (63) and where Romanian is the most common main language other than English (45); the fourth colours each district by its education authority’s 2026 pupil count (see Figure D3). Hover any authority for counts and for Romanian’s rank among the languages spoken there. Source: ONS TS024, custom extract; ONS Census 2021 main language by local authority; DfE FOI-2026-0093642 (fourth view); author’s analysis.

    Therefore, concentration is pronounced with the ten largest authorities accounting for 28% of all Romanian speakers, while the twenty largest hold nearly 40%. This geographical distribution also has some practical implications. Because two-fifths of the Romanian-speaking population resides in some twenty local authorities, a small network of teaching centres and examination venues (i.e.: a dozen well-placed locations) could potentially serve a large share of prospective students/candidates.

    Figure B2
    How few authorities hold how many speakers
    Cumulative share of all Romanian main-language speakers in England and Wales, local authorities ranked from largest to smallest Romanian population. Hover the curve to read any rank. Source: ONS TS024, custom extract (331 authorities, 471,952 speakers); author’s calculations.

    The remaining three-fifths, however, are dispersed in more than three hundred local authorities, in numbers perhaps too small to sustain local provision. For them, the established solution, without incurring exorbitant public resources, is the private-candidate route: pupils prepare in community or supplementary schools, or even at home, and sit the examination at a registered centre that accepts external entrants.14 Both Polish and Arabic-language entries use this model, so a Romanian qualification would require no new or improvised mechanism, except for the two-track structure that existing community languages already use.

    From Thurrock and the eastern boroughs, through Luton and Milton Keynes, to the two Northamptonshire authorities, this corridor is in fact, contiguous, so that (by way of problem-solving) neighbouring concentrations can share educational provision. A teaching or examination venue located along this corridor may draw candidates from several local authorities at once, which lowers the numbers of sites a viable network requires. In this sense, the comparison with Polish settlement could be instructive. Romanian speakers now outnumber Polish speakers in 63 of the 331 authorities (LAs), including the leading boroughs, by more than six to one in Harrow and Redbridge.

    Figure B3
    Where Romanian leads Polish: speakers of each, the fifteen widest margins
    Romanian and Polish main-language speakers in the fifteen authorities where the ratio of Romanian to Polish speakers is highest, ranked by that ratio (shown on the right). Source: ONS TS024, custom extract; author’s calculations.

    Polish sustains the largest community-language entry in the country – 8,419 GCSE candidates in 2024 – from a base even more dispersed across the administrative geography of local authorities. Correspondingly, a Romanian qualification would recruit from a population that is larger in its core locations and easier to reach with fewer centres.

    CSection C

    Brief overview of community-languages provision & certain anomalies

    Setting the census speaker populations against summer 2024 examination entries yields the following: Romanian, with 471,954 speakers, has no qualification and no entries.

    Panjabi (290,745 speakers) recorded 964 entries, Urdu (269,849) 4,182, Portuguese (224,719) 2,734, Arabic (203,998) 5,256, and Italian (160,010) 4,176. Expressed as a rate, the existing community-language qualifications draw between roughly 12 and 26 GCSE entries per thousand speakers (Portuguese 12.2, Polish 13.8, Urdu 15.5, Turkish 21.4, Arabic 25.8), with Panjabi (3.3), where language appears to have thinned furthest by census reporting, the sole outlier below.

    Figure C1
    Speaker population against GCSE entries, with the implied Romanian entry range
    Apply the entry rate of
    Main-language speakers (Census 2021, England and Wales) against GCSE entries (summer 2024, UK, all boards) for every community language currently examined. Curricular French, German and Spanish are excluded, as their entries derive from taught provision rather than heritage-speaker demand. The orange bar shows the entry Romanian’s 471,954 speakers would imply at the observed rates of the comparable community languages (5,742 at the Portuguese rate to 12,160 at the Arabic rate); the dashed line marks the circa 5,000-candidate figure cited in the parliamentary petition record (see note 15). Illustrative standardisation, not a forecast. Sources: ONS, Census 2021, Figure 1 data download; JCQ, GCSE Other MFL entries, summer 2024; author’s calculations.

    Applied illustratively to the Romanian speaker base, these rates imply an entry range of approximately 5,700 to 12,200 at maturity; that is, even the least engaged of the comparable community-language rates would comfortably surpass the circa 5,000-candidate threshold that has been cited in at least one previous instance as a potential viability benchmark,15 and the upper half of the range would place Romanian second only to Polish among community languages.

    Even the least engaged comparator rate implies c. 5,700 Romanian entries, above the c. 5,000-candidate benchmark.Portuguese rate, 12.2 entries per 1,000 speakers, applied to 471,954 speakers.

    As previously noted, two properties of the Romanian population suggest the range is conservative: its age structure is concentrated at family-forming ages to a degree unmatched by any comparator community, so its school-age population is still growing into the educational system; and the main-language measure excludes precisely the bilingual second generation from which heritage-language entries disproportionately come.

    Figure C2
    The existing community-language GCSEs are growing, not static
    GCSE entries in Polish and Arabic, summer series, UK, all awarding bodies. 2020 and 2021 are omitted (no summer examinations; centre- and teacher-assessed grades); Polish 2025 not yet in the compiled series. Source: JCQ, GCSE (full course) entries, Other MFL.

    The Department’s (DfE’s) pupil counts, obtained after this estimate was built (section D), allow it to be checked from the opposite direction. The estimate above starts from speakers in the census; the check commences from pupils in schools. There are 5,316 pupils with Romanian first language in Year 11 right now – above the cited benchmark before any modelling at all. We can also observe how such a cohort behaves where a qualification is already in place: Polish GCSE entries run at roughly eight for every ten pupils in the Polish Year 11 cohort. At that same propensity, the Romanian-language current cohort would yield around 4,200 entries immediately – rising every year, since the year groups behind it are larger still.

    Two independent measures – census speakers and enrolled pupils – converge on the same scale of demand, understood here in the examination sense: the number of candidates who could realistically be expected to enter each year, judged by how comparable communities behave where a qualification exists. Thus, by either route, a Romanian GCSE draws entries in the thousands each year from its first sitting, at or above the circa 5,000 benchmark.16

    DSection D

    The school layer – the pupils counted

    The Department for Education records every pupil’s first language in the January school census, however, it publishes the result only in aggregate: pupils whose first language is English, pupils whose first language is other than English – the category known as EAL (English as an additional language) – and unclassified.

    No named languages appear in the published series, so EAL figures cannot identify Romanian pupils directly. For this report, the named-language counts were obtained directly: headcounts of pupils recorded with Romanian, and as comparator Polish, as a first language, by National Curriculum year group and local authority, from the January school censuses of 2016, 2021 and 2026.17 Before these figures were secured, the published EAL aggregates served this analysis as an indirect indicator of arrival and as an upper bound on possible numbers. The counts now verify both: every figure below sits within its EAL ceiling, and the places the aggregates initially singled out, Doncaster above all, are precisely where the counted populations proved largest.18

    Figure D1
    Pupils with a first language other than English, selected local authorities, 2015/16–2025/26
    Percentage of pupils across all phases whose first language is recorded as other than English (EAL). The aggregate signal through which the arrival was first traced, before the named-language counts were obtained: Doncaster’s share nearly doubled (1.8), rising from 8.6 to 15.5 per cent, and accelerating with the post-2014 window of Romanian settlement; the mature London boroughs show stable or gently declining shares, consistent with change in composition rather than scale – Newham declining gently to 65.6, Harrow plateauing at 64.4. Source: DfE, Schools, pupils and their characteristics, 2015/16–2025/26, named counts, DfE FOI-2026-0093642; author’s analysis.

    Nationally, England’s schools held 98,210 pupils with Romanian as first language in 2025/26 – a tripling in a decade, from 32,143 in 2015/16 – of whom 28,306 are at secondary phase, and 5,316 in Year 11, the GCSE year (2026). The cohort structure matters as much as the total. From Year 10 down to Year 1, every younger year group is larger than the one above it, from 5,387 in Year 10 to more than 10,000 in each of Reception, Year 1 and Year 2, so the GCSE-age cohort roughly doubles over the coming decade on enrolment alone, before any further arrival and settlement. The Polish language comparator, also evidences this transition: Polish pupil numbers peaked in 2020/21 at 140,170 and have since declined to 126,092, and from Reception to Year 3, Romanian pupils now outnumber Polish in every year group. Thus, in England’s youngest classrooms, Romanian is already the largest first language after English.19

    Figure D2
    Pupils with Romanian or Polish as first language, by National Curriculum year group, England, state-funded schools, January 2026, with each group’s January 2016 headcount marked with a black dot
    From Year 10 down to Year 1, every Romanian year group is larger than the one above it (Reception, 10,083, is just below Year 1, 10,429); from Reception (R) to Year 3, Romanian language pupils exceed Polish, the largest community first language nationally. Compulsory year groups shown. Source: DfE, FOI-2026-0093642. Hover a year group for the 2016, 2021 and 2026 counts.

    The pupil geography confirms the census map, and completes the landscape. Harrow leads among the lower-tier authorities – 4,059 pupils, 1,093 at secondary phase, 211 in Year 11 alone. It is followed by West Northamptonshire (3,859), and Birmingham (3,174); Doncaster, whose EAL trajectory first signalled the arrival, holds 1,112 pupils with Romanian as first language. A new finding emerges from the granular data obtained from DfE. Hertfordshire, invisible in the district-level census rankings, holds the largest Romanian pupil population of any education authority in England (4,070), with Essex, Kent and Leeds each above 2,000, mapping onto the corridors previously identified.20

    Figure D3
    Pupils with Romanian as a first language, by education authority, England, January 2026
    Largest Romanian pupil populations, January 2026
      The school census’s upper-tier geography reveals what census districts disperse: Hertfordshire (4,070) holds the largest Romanian pupil population of any education authority in England, and the settlement corridor runs continuously from the Thames estuary through Hertfordshire to Northamptonshire. Harrow (4,059; 1,093 at secondary phase) remains first among boroughs; Essex, Kent and Leeds each exceed 2,000 pupils. Colour scale: square root, to keep mid-sized authorities distinguishable; Wales lies outside the extract’s scope. Source: DfE, FOI-2026-0093642; boundaries: ONS Open Geography Portal, census districts dissolved to education authorities by the authors. Hover an authority, or a name in the list, for exact counts.

      The growth has also redistributed. In 2016, nearly half of all Romanian pupils were in London schools; today the capital holds 28.5 per cent – not because London contracted (its count in fact doubled). The rest of England grew faster still: nearly fivefold in the East Midlands (2,162 to 10,363 pupils), more than fourfold in the East of England (now the second region, with 16,407) and nearly fourfold in the South East, with every one of England’s nine regions holding at least 1,700 Romanian pupils. This has already shaped into a national-level school population with a London arc.21

      The decade’s trajectory is also instructive. In January 2016, no Romanian year group in the country reached five thousand pupils – the largest held 4,138. By 2021, eight of the twelve compulsory year groups had passed the mark; by 2026 all twelve have. The Year 11 cohort has quadrupled across the three school and pupil censuses (1,201 to 3,770 to 5,316). Because these cohorts advance one year group each year, the enrolment registers already contain the Year 11 cohorts of the next decade: 5,914 within three years, 7,780 within five, approaching 9,000 by the mid-2030s (after the fifteen per cent easing observed as cohorts enter secondary phase). Nor did this growth depend on hard borders and free movement restrictions. Pupil numbers rose 124 per cent in the five years to 2021, and a further 36 per cent in the five years after free movement ended, post-Brexit. The later rise was principally driven by children already in the UK, aging into the British school system.22

      Setting these counts against the census age and main language table commissioned in parallel, yields another consequential measure. Pupils recorded with Romanian as a first language outnumber the census count of school-age Romanian speakers in every comparable authority, by a factor of 1.8 to 3.6 across the eighteen priority authorities – 2.0 in Harrow, 2.4 in West Northamptonshire, 2.2 in Doncaster, a median of 2.2 and 2.2 nationally (98,210 pupils against 44,323 census-measured children). This multiplier captures the bilingual second generation – children raised with Romanian at home, who report, or are recorded with, English – whom the census main-language question cannot count. In effect, the true school population runs at roughly twice the census count, and nowhere at less than 1.75 times it.23

      Figure D4
      Two counts of the same children
      Census 2021 school-age Romanian main-language speakers (ages 3-14) against pupils recorded with Romanian as first language, January 2026, by local authority. The counted school population exceeds the census measure everywhere, by a factor of 1.8 to 3.6 (2.2 nationally) – the bilingual second generation invisible to a single-response language question, together with five years of arrivals. Sources: ONS CT21_0503; DfE FOI-2026-0093642; authors’ calculations (ratio = pupils ÷ census school-age). Eighteen priority authorities, each a unitary authority or London borough, identical in both geographies. Hover a row for exact counts.
      ESection E

      The Romanian share of the school population – potential avenues for provision & delivery

      One further measure can be constructed and it asks a simple question directly: within each local authority’s multilingual school population, how large is the Romanian presence? The measure represents the share of secondary-phase EAL pupils whose first language is Romanian – both counts from the same January 2026 school census, so a true percentage.24

      The purpose of this measure is strategic, aimed at the educational delivery decisions the raw counts do not necessarily illuminate. Whether a Romanian GCSE exists at all rests with an awarding organisation, nationally; but once it exists, whether it is taught, and where it is sat, would be decided at school or examination centre-level. In this instance, relative weight matters more than absolute size: where the share is high, Romanian pupils form a substantial presence within a small and comparatively undiversified multilingual school population; where it is low, they sit inside a large school population drawn from many language communities, the conditions under which provision has historically run through supplementary schools. Measured by population alone, Birmingham (with 10,573 Romanian speakers on the ONS 2021 census) ranks above Doncaster (with 5,428). Yet for a school wondering whether to offer Romanian, Doncaster would qualify as a stronger environment, because its pupils are a far larger presence within a far smaller multilingual pupil body. However, there are other considerations impacting such decisions, a pre-existing educational infrastructure stemming from the community itself, a proportion funded by the Romanian government through its diaspora outreach mechanisms – some of these can already be mapped, whether parish weekend schools or more informal educational initiatives from community associations. We shall endeavour to map some of these based on the Romanian funding distribution abroad in a following section.

      Figure E1
      Romanian share of the multilingual secondary school population, by priority authority
      Rank by
      Pupils with Romanian as first language as a percentage of pupils with a first language other than English, state-funded secondary schools, January 2026. The settlement-corridor authorities occupy the entire upper tier (16.6 to 11.7 per cent), favouring teaching and examination within mainstream host schools; in the lower tier (2–3 per cent), substantial Romanian populations sit within very large, highly diverse school populations — the environment in which community-language provision has historically operated through supplementary schools and private-candidate entry. Source: DfE, FOI-2026-0093642 and Schools, pupils and their characteristics 2025/26; authors’ calculations. Switch to “Speaker count” to rank the same authorities on population alone, or to the retired intensity index of earlier editions. Hover a bar for exact counts.

      The measured shares of pupils with Romanian as first language, divide the eighteen priority authorities in two clear tiers. At the top, the Thurrock-Northamptonshire corridor, joined by its east-London end: West Northamptonshire 16.6 per cent, Thurrock 14.9, North Northamptonshire 14 per cent, Havering 13.5, Doncaster 11.7 per cent – authorities where roughly one multilingual secondary pupil in six or seven speaks Romanian, where Romanian is the principal community language of the local school system, and where mainstream schools could credibly choose to teach and examine the subject. At the bottom, the large diverse cities and local/educational authorities: Enfield and Birmingham at 2.8 per cent, Slough at 2.3, where substantial Romanian populations are diluted within very large school populations drawn from dozens of communities, the environment in which Polish and Arabic provision have in practice operated through supplementary-school and private-candidate routes described in section B. The share, in other words, maps onto the two delivery models without predicating where provision will in fact emerge; it identifies the authorities where each model is demographically viable, while the decisions remain at the latitude of competent bodies. This should be regarded as an instrument of readiness: it shows an awarding organisation, in advance, through what type of provision candidates could plausibly reach examination – and this, in and of itself, constitutes the delivery side of the business case.

      Birmingham: c. 10 Romanian secondary pupils per secondary school if spread evenly. Harrow: c. 84.926 pupils across 91 secondaries against 1,093 across 13; DfE FOI-2026-0093642 and school-level data, 2025/26.

      The school infrastructure reinforces this argument. In Birmingham, 926 Romanian-first-language secondary pupils are dispersed in 91 state-funded secondaries – around ten per school if spread evenly – so provision for any single language is a matter of accumulating critical mass. In West Northamptonshire, 1,066 secondary pupils map onto 23 schools, around 46 per school; in Harrow, 1,093 onto 13, some 84 per school. Since families cluster rather than spread evenly, concentrations in particular schools may sit well above these floors. In the local authorities spanning the geographical corridor, an individual secondary could realistically assemble a class of timetable-viable size.25

      Harrow remains the instructive exception: first in absolute numbers and population share, it still carries a 10.7 per cent share of one of the most linguistically diverse school populations in England – implying that both delivery models could be viable simultaneously.

      FSection F

      Age structure and the first-examination cohort

      Figure F1
      Age profile of speakers, selected community languages, Census 2021
      Percentage of each language’s main language speakers in each age band. Source: ONS, commissioned table CT21_0503 (national profiles summed from local authority counts); author’s calculations. Hover the chart for each band.

      Several dimensions flow from the more granular age-based appraisal. Firstly, the school-age population is already significant, with 44,323 Romanian-speaking children aged 3 to 14 years, more than any language community except Polish. Secondly, it is still growing within the educational system: children aged 3 to 9 (27,895) outnumber those aged 10 to 14 (16,428) by 1.7 to one, and the younger group reaches Year 11 (also the GCSE year) between roughly 2027/28 and 2033/34 – the window in which a new qualification would hold its first examination.26 On the census count alone, this represents a cohort of about 4,000 pupils a year, rising across the period. Because the table was commissioned at local-authority level, the cohort can also be placed: Harrow holds 1,993 school-age Romanian speakers on the census measure, Birmingham 1,734, West Northamptonshire 1,580 – and the same age structure, younger cohorts outnumbering older, repeats in every priority authority, so the pipeline rises everywhere at once.27

      Figure F2
      From census to Year 11: Romanian-speaking children aged 3–9 in March 2021 reach Year 11 across 2027/28–2033/34
      0of 27,895
      Illustrative: the 27,895 children aged 3–9 recorded at census, spread evenly (c. 4,000 a year) across the seven Year 11 cohorts they will enter. Not a forecast; migration, attrition and school-system entry are not modelled. Source: ONS, commissioned table CT21_0503.

      Thirdly, it is important to note that these figures may yet be conservative estimates, since the census records only a child’s main language, so UK-born children of Romanian-speaking homes who report English have not been yet included, while the count itself predates five years of further arrivals and births. The school data presented in section D have since measured this understatement directly: pupils recorded with Romanian as a first language outnumber the census’s school-age speakers by a factor of 2.2 nationally, and by 1.8 to 3.6 in every comparable authority. The Home Office’s settlement data illustrate the same gap from a third direction. By June 2026, 137,101 Romanian children under 18 held settled status (roughly three times the census count of school-age Romanian speakers), with a further 116,649 holding pre-settled status, based on the count of granted applications. While these data are not as granular as the census, they help evidence the Romanian community through a different lens, counterbalancing the more conservative measures that may underestimate the real population size.28

      Figure F3
      Three measures of the Romanian child population
      School-age Romanian main-language speakers (Census 2021, England and Wales, ages 3-14) against Romanian children under 18 granted pre-settled and settled status under the EU Settlement Scheme (UK, cumulative to June 2026, by latest case outcome). The instruments differ in measure, geography and time range; it is illustrative however, of a potentially bilingual second generation imperfectly captured, and should not be treated as a like-for-like count. Sources: ONS, CT21_0503; Home Office, EUSS detailed tables, June 2026; author’s calculations.

      One further figure from the 2021 census table addresses the question of proficiency in English-language. To this end, 81.3% of Romanian main-language speakers report speaking English well or very well, showing a community striving to communicate in English while maintaining Romanian, the language of heritage. This bilingual profile also ought to be diligently recognised by the qualifications’ boards.29

      GSection G

      A county-level validation: Hampshire

      Contributor data from Hampshire’s Ethnic Minority and Traveller Achievement Service (EMTAS) provide an independent, service-level check on the national extract presented in section D – pupil counts assembled by a county service for its own operational purposes, from its own Spring census returns (independent of the DfE freedom of information request). The two instruments agree to within one per cent: EMTAS recorded 1,095 pupils with Romanian as a first language in Hampshire’s schools in spring 2025; the Department’s national extract returns 1,108 for 2025/26. Three findings reach beyond the validation itself.

      Section G draws on contributor-level data supplied by Hampshire EMTAS: service records and Spring Census extracts shared with the authors, not published official statistics. It is presented as a local validation of the national picture, not as a representative sample.

      The first concerns scale in an unexpected place. Hampshire is not necessarily a priority authority (the Romanian community is a modest share of a large county). Yet its schools hold more than a thousand pupils, and the number is moving: up ten per cent in two years, from 994 to 1,095, over exactly the period in which Polish pupils’ numbers fell four per cent, from 2,367 to 2,265. The transition of the generations that section D measured nationally is visible here more granularly, on a single measure – and if a county quite removed from the community’s heartland holds a thousand pupils, the dispersed three-fifths of the population described in section B is anything but negligible.

      Figure G1
      Pupils by first language, Hampshire, all year groups, all schools, 2023–2025
      Pupils whose first language is Arabic, Polish or Romanian, all year groups, all Hampshire schools. Source: Spring School Census, via Hampshire EMTAS (contributor data).

      The second finding concerns demand. A referral is a pupil whose school has asked the county’s specialist service for support – most often a newly arrived child requiring language assessment or classroom help. Romanian referrals have run level with Polish across three years, but from a pupil population half the size. Relative to its numbers, the Romanian community is asking roughly twice the support.

      The third finding concerns delivery, and it introduces an actor this report has not yet named. The same service shepherded 88 Polish pupils through their heritage-language GCSE in 2026, up from 63 in 2023 – proof that where support exists, entries grow, and that county services of this kind amount to a third route to examination, alongside the host and supplementary schools of section E. The Polish figures allow an informative illustration: 88 supported entries from 2,265 Polish pupils is an engagement rate of roughly four per cent a year. At that same rate, Hampshire’s thousand-plus Romanian pupils would yield around forty entries annually – from just one county with no Romanian concentration to speak of.

      Figure G2
      Heritage-language GCSE entries supported by Hampshire EMTAS, 2023–2026
      Number of heritage-language GCSE candidates supported by Hampshire EMTAS, by language and examination year, 2023–2026. Romanian is shown empty because no Romanian GCSE exists to sit. Source: Hampshire EMTAS (contributor data).
      Figure G3
      Referrals to Hampshire EMTAS, 2023–2025
      Pupils referred to the Ethnic Minority and Traveller Achievement Service, by first language, 2023–2025. A referral is a pupil whose school has asked the county’s specialist service for support – most often a newly arrived child requiring language assessment or classroom help. Source: Hampshire EMTAS, own data (contributor-level data).
      HSection H

      Romanian state funding for Romanian-language education in the UK

      The delivery environments briefly covered in section E indicate whether a community teaching infrastructure already exists, and more specifically if there is any meaningful demand for the transmission of Romanian language, beyond the home/domestic setting. For the Romanian community, part of this infrastructure is directly observable through an unusual instrument – the grant records of the Romanian government’s Department for Romanians Abroad (Departamentul pentru Românii de Pretutindeni, DPRP), audited at project-level by the civic transparency initiative iGov, and The Diaspora Initiative. Between 2017 and 2024, DPRP awarded grants for organisations registered in the United Kingdom for Romanian-language education and community activity; each record names the recipient, the project and the funds received. This may also be read as an administrative census of the Romanian community’s own (informal) school network.

      The record shows both appetite and demand for the transmission of Romanian language amongst children and pupils, this time through community-level and informal education initiatives – weekend schools, parish classes, courses and camps, run by community associations and congregation outside any state curriculum. We can surmise, that outside any formal qualification, performance credit or institutional incentive, enrolment in such grassroots initiatives has been motivated by the families’ own commitment to transmitting the language to their children, as they volunteer their Saturdays and sustain these schools year after year. The Romanian government underwrites part of this effort, through its diaspora educational axis.

      This section uses iGov’s audit of the grants awarded by the Romanian government’s Department for Romanians Abroad (Departamentul pentru Românii de Pretutindeni, DPRP) to organisations registered in the United Kingdom, 2017–2024. Amounts are in Romanian lei (RON) as recorded: approved is the grant awarded, settled the amount accounted for and paid out. Projects are classified from their titles: Romanian-language schools are teaching projects for a named Romanian school (the “Nicolae Iorga” network and others); parish schools are schools run by a church; other education covers courses, camps, digital resources and events in DPRP’s education categories.

      Figure H1
      DPRP funding settled with UK organisations, by year and purpose (RON)
      Settled amounts, all UK grants, 2017–2024, stacked by purpose. Hover a bar for values. Source: DPRP grant records, compiled and classified by iGov (dprp.igov.ro); author’s classification of school projects.
      Figure H2
      Every school grant, 2017–2024: where, who, how much
      Each line is one grant to a Romanian-language school or a parish school. The pale bar is the amount approved, the solid bar the amount settled. Locations in Northern Ireland are marked (NI): those pupils would sit CCEA, not English, GCSEs. Source: DPRP grant records via iGov (dprp.igov.ro).
      Figure H3
      All UK funding by type of organisation, and what it paid for (RON, settled)
      Settled amounts 2017–2024 by the organisation type recorded by DPRP, split by purpose. Church funding is dominated by three property purchases: St Paul’s Centre, Enfield (2023), a place of worship in Plymouth (2024) and a church and cemetery in Swansea (2024). Hover a segment for detail. Source: DPRP grant records via iGov (dprp.igov.ro).

      Therefore, what a Romanian examination would add is an endpoint, premised on formal recognition. The language is already being taught at the community’s own expense. A GCSE would convert this sustained private effort, into publicly certified attainment.

      Heritage-language GCSEs count in full within the English Baccalaureate’s languages pillar, on the same terms as French, German, and Spanish – a parity the Government has confirmed on the parliamentary record, noting that these languages ‘all count towards the English Baccalaureate’s (Ebacc’s) school performance measure. The incentive, therefore, runs in both directions: for the pupil, a recognised grade on a national certificate – understood by any sixth form, university or employer in the country; for a school that hosts the teaching and enters the candidates, a direct contribution to its own Ebacc language figures – one reason heritage-speaker provision has proved attractive to host schools, and a mechanism a Romanian GCSE would inherit once approved by the examination board.

      To conclude

      This body of research is still fluid, as the distribution of pupils declaring Romanian as first language evolves, graduating year by year, as we have shown, in ever larger cohorts. Each January census from the Department for Education will redraw the figures and maps presented here, but not necessarily the spine of our case – that Romanians shaped into a consequential and strategic constituency, their children already embedded in the UK’s educational system, its school population approaching 100,000 nationally. Language transmission is poignant, Romanian is taught informally through the myriad community initiatives sustained through grassroots efforts, largely voluntary, year by year. Romanian is also spoken at home, feeding the personal networks that extend into the country of heritage. There are manifold implications for the viability of a Romanian-language GCSE, but the most important aspect is its absence, and the disenfranchisement this vacuum perpetuates for the pupils themselves. Since 2019, petitions to the UK Parliament, reiterated in 2026, together with substantial correspondence with the Education Secretaries under changing governments, as well as community-level initiatives, did not yield any meaningful progress.

      We are, therefore, presenting a body of evidence that aims to shift the knowledge base regarding Romanian communities, while informing policy and provision well beyond the examination itself – school place planning, language support services, county level services accessed by the schools themselves. This should also be read as business case. The questions an awarding organisation must answer before introducing a qualification – is there demand, where is it, can it be delivered, will entries sustain the examination – are answered here in sequence. Until then, the largest community language to arrive in a generation remains the largest without a qualification, from any awarding body.

      The largest community language to arrive in a generation remains the largest without a qualification, from any awarding body.

      Sources

      1. Office for National Statistics, Language, England and Wales: Census 2021 (statistical bulletin), Figure 1 and Figure 3 data downloads. www.ons.gov.uk/peoplepopulationandcommunity/culturalidentity/language/bulletins/languageenglandandwales/census2021
      2. Office for National Statistics, Census 2021: TS024 Main language (detailed), lower-tier local authorities, via Create a Custom Dataset. www.ons.gov.uk/datasets/create
      3. Office for National Statistics, Census 2021: CT21_0418, Main language (detailed) by proficiency in English by age, England and Wales.
      4. Office for National Statistics, Census 2021: country of birth and national identity, England and Wales.
      5. ONS Open Geography Portal, Local Authority Districts (December 2021) boundaries. geoportal.statistics.gov.uk
      6. Joint Council for Qualifications, GCSE (full course) entries and results, Other MFL, summer 2019–2025. www.jcq.org.uk/examination-results/
      7. Ofqual, provisional entries for GCSE, AS and A level, summer 2025 (England).
      8. Department for Education, Schools, pupils and their characteristics, 2015/16 to 2025/26. explore-education-statistics.service.gov.uk/find-statistics/school-pupils-and-their-characteristics
      9. Home Office, Immigration system statistics, year ending June 2026: EU Settlement Scheme detailed tables (EUSS_D01, EUSS_D02). www.gov.uk/government/statistical-data-sets/immigration-system-statistics-data-tables
      10. iGov, audit of DPRP (Departamentul pentru Românii de Pretutindeni) grants 2017–2024, project-level records. dprp.igov.ro
      11. Department for Education, response to freedom of information request FOI-2026-0093642: pupils with first language recorded as Romanian or Polish, by National Curriculum year group and local authority, January 2016, 2021 and 2026.
      12. Office for National Statistics, Census 2021: commissioned table CT21_0503, Main language (detailed, 67 categories) by age, lower-tier local authorities, England and Wales (produced for this report, 15 October 2026).
      13. ONS Open Geography Portal, Lower Tier Local Authority to Upper Tier Local Authority (December 2021) lookup in England and Wales.
      14. Home Office, EU Settlement Scheme quarterly statistics, June 2023: repeat applicants (EUSS_RA_01–04); and Immigration system statistics, year ending June 2026: citizenship grants.
      15. Office for National Statistics, Births by parents’ country of birth, England and Wales; Long-term international migration, provisional, year ending December 2025.
      16. Hampshire Ethnic Minority and Traveller Achievement Service (EMTAS): heritage-language GCSE support 2023–2026, Spring Census pupil extracts 2023–2025 and EMTAS referrals 2023–2025 (contributor data).

      Footnotes

      1. Home Office, Immigration system statistics, year ending June 2026, EU Settlement Scheme detailed tables (EUSS_D01, applications; EUSS_D02, concluded applications by latest case outcome), nationality = Romania: 1,998,079 applications; 686,129 concluded in settled status, 726,404 in pre-settled status. Available at: www.gov.uk/government/statistical-data-sets/immigration-system-statistics-data-tables. Repeat applicants are identified by the Home Office’s data-matching across applications: Home Office, EU Settlement Scheme quarterly statistics, June 2026 (tables EUSS_RA_01–04, 28 August 2018 to 30 June 2026): 532,200 Romanian repeat applicants, 31 per cent of all EU-national repeat applicants (1,741,410) and the highest of any nationality; 376,280 of them now hold settled status, 348,510 having moved from pre-settled to settled status. At 30 June 2023 the count was 309,680 (29 per cent); that release is available at: www.gov.uk/government/statistics/eu-settlement-scheme-quarterly-statistics-june-2023/eu-settlement-scheme-quarterly-statistics-june-2023. The unique-persons figures are the authors’ netting of repeat applicants from application totals; the Home Office publishes no unique-persons count by nationality. ↩
      2. ONS, Long-term international migration, provisional: year ending December 2025 (administrative-based estimates; the latest periods are provisional and subject to scheduled revision): Romanian the most common EU+ nationality among both long-term immigrants and long-term emigrants. Available at: www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/bulletins/longterminternationalmigrationprovisional/yearendingdecember2025. ↩
      3. Citizenship: Home Office, Immigration system statistics, year ending June 2026, “How many citizenship grants have been issued in the UK?” (YE June 2026: 8,285 grants to Romanian nationals, the second-largest EU nationality after Italian; naturalisation includes individuals previously granted EUSS settled status): www.gov.uk/government/statistics/immigration-system-statistics-year-ending-june-2026/how-many-citizenship-grants-have-been-issued-in-the-uk. Births: ONS, Births by parents’ country of birth, England and Wales (annual bulletins; dataset: Parents’ country of birth): series points 2007: 1,248; 2017: 13,717; 2018: 15,196; 2019: 16,069; most common non-UK country of birth of mothers in 2021, second after India from 2022. Bulletin: www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/livebirths/bulletins/parentscountryofbirthenglandandwales/2019; dataset: www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/livebirths/datasets/parentscountryofbirth ↩
      4. DfE, response to FOI-2026-0093642: pupils with first language recorded as Romanian, state-funded schools, England, January 2026 school census; presented in full in section D. ↩
      5. July 2015 government intervention, Future of community language qualifications secured (DfE official press release). Available online at: www.gov.uk/government/news/future-of-community-language-qualifications-secured. This was followed by the April 2016 settlement, Community languages saved to ensure diverse curriculum continues (DfE official press release, includes the languages, the Pearson/AQA reallocation, and the Ofqual-brokered agreement). Available online at: www.gov.uk/government/news/community-languages-saved-to-ensure-diverse-curriculum-continues. Also refer to, DfE (December 2023). British Sign Language GCSE subject content, available online at: www.gov.uk/government/publications/gcse-british-sign-language. ↩
      6. Community-language GCSEs count in EBacc languages pillar on equal terms – DfE, English Baccalaureate (EBacc) guidance (the languages pillar admits any approved GCSE language, ancient or modern), resource available online at: www.gov.uk/government/publications/english-baccalaureate-ebacc/english-baccalaureate-ebacc. ↩
      7. Census 2021 applies targeted record swapping and cell-key perturbation (ONS, Protecting personal data in Census 2021 results). DfE ad hoc extracts conventionally suppress small counts (marked “c”), apply secondary suppression to adjacent cells to prevent back-calculation, and round remaining values to the nearest five; the extract has been asked to state the threshold applied. The aggregate extract analysed in the following sections contains no suppressed cells. Suppression was expected in the named-language extract; in the event, the Department supplied unrounded headcounts by year group and authority (FOI-2026-0093642), and no bounding was required. ↩
      8. Any extension of the analysis into shire counties requires aggregation of census districts to county-level. The named-language information request will be specified at the 153-authority geography accordingly. ↩
      9. Peach, C. (2006) ’South Asian migration and settlement in Great Britain, 1951–2001’, Contemporary South Asia, 15(2), 133–146. ↩
      10. Linguistic Minorities Project (1985) The Other Languages of England (Routledge 1st edition, revised in 2018); Harris, R. (2006) New Ethnicities and Language Use (Palgrave Macmillan). Passim.; Harris charts how everyday interactions shape the linguistic repertoire and rely on a rich mix of London English, heritage languages and elements of Jamaican and African American expressions. ↩
      11. The ONS Language, England and Wales: Census 2021 Bulletin, stating that the Romanian language increase mirrors the increases in Romania as country of birth and Romanian national identity. Available online at: www.ons.gov.uk/peoplepopulationandcommunity/culturalidentity/language/bulletins/languageenglandandwales/census2021. ↩
      12. Home Office, Immigration system statistics, year ending June 2026, EUSS detailed tables (EUSS_D01/D02): www.gov.uk/government/statistical-data-sets/immigration-system-statistics-data-tables. Author’s extraction, nationality = Romania. Counts are of applications, not unique individuals: repeat submissions are common (including pre-settled holders reapplying for settled status), so neither the application total nor any sum across outcomes equals a number of people. The settled-status figure is the more conservative reading. ↩
      13. DfE, English proficiency of pupils with English as an additional language (2020). Available online at: www.gov.uk/government/publications/english-proficiency-pupils-with-english-as-additional-language. ↩
      14. DfE, answer to written question HL14429 (Baroness Bennett), 7 April 2021 – acknowledging that community-language candidates typically study outside mainstream schools and enter as private candidates: questions-statements.parliament.uk/written-questions/detail/2021-03-22/HL14429/. See also ’Qualifications in home languages’, JMMD (open access): doi.org/10.1080/01434632.2023.2288292; Global Future, Silenced Voices (2021): ourglobalfuture.com/reports/silenced-voices/. For Polish-language instruction, please refer to, Polska Macierz Szkolna examination centre – polskamacierz.org/. ↩
      15. No formal entry threshold is published by any awarding organisation; a figure of circa 5,000 candidates circulates in the parliamentary petition record as the boards’ business-case benchmark, while the Government’s stated position is that introduction rests at the discretion of exam boards, which it would support. UK Parliament petition 617718, Add Romanian to the Modern Languages GCSE (2022; 11,000+ signatures), Government response, 22 July 2022: “There is no reason in principle why a GCSE in Romanian could not be introduced, and the Government would support any exam board wishing to do so. This is, however, at the discretion of exam boards…’ There are currently nineteen GCSEs in different modern languages produced and currently offered by exam boards in England.” Available online at: petition.parliament.uk/petitions/617718. ↩
      16. Author’s calculation from FOI-2026-0093642: Polish GCSE entries, summer 2024 (8,419; JCQ), set against the Polish Year 11 cohort of roughly 10,600 pupils – a ratio of about 0.8. Entries include some candidates outside Year 11, so the ratio slightly overstates Year 11 engagement; applied here to Romanian Year 11 cohort of 5,316 as an illustration, not a forecast. ↩
      17. DfE, response to FOI-2026-0093642: pupils with first language recorded as Romanian (codes RMN, RMNM, RMNR) and Polish (POL), state-funded schools, England, by NC year group nationally and by local authority; January censuses 2016, 2021 and 2026; headcounts supplied unrounded. ↩
      18. Verification details: secondary-phase counts against 2025/26 secondary EAL ceilings – Harrow 1,093 of 10,184; West Northamptonshire 1,066 of 6,433; Birmingham 926 of 32,860; Newham 780 of 16,803. Doncaster: multilingual share 8.6% to 15.5% across the decade, acceleration post-2014; 1,112 Romanian-first-language pupils counted in 2025/26. ↩
      19. Same source, national tables. Secondary phase = Years 7–11 (6,164 + 5,914 + 5,525 + 5,387 + 5,316). Reception-to-Year-3, 2025/26: Romanian 10,083 / 10,429 / 10,416 / 10,067 against Polish 7,434 / 8,305 / 9,097 / 9,838; the superlative follows from Polish being the largest community first language nationally. ↩
      20. Upper-tier geography: shire counties appear here as single education authorities but as multiple districts in census outputs, so these totals have no single-district census counterpart (see the Data section on the two geographies). ↩
      21. DfE, FOI-2026-0093642, regional sums of authority counts; authors’ allocation of education authorities to regions. Regional totals, 2025/26: London 27,985; East of England 16,407; South East 12,210; West Midlands 11,723; East Midlands 10,363; Yorkshire and The Humber 7,189; North West 5,909; South West 4,633; North East 1,791. London’s 2015/16 share: 47 per cent of mapped pupils, and no lower than 44 per cent of the national total (authorities reorganised since 2016, c. 6 per cent of that year’s total, are excluded from the 2015/16 regional sums; the 2025/26 allocation is complete). The East Midlands comparison includes Northamptonshire (1,243 pupils in 2015/16), the predecessor of North and West Northamptonshire. ↩
      22. Same source, national tables. Year groups at or above 5,000 pupils: none of twelve in 2015/16 (maximum 4,138, Year 1); eight of twelve in 2020/21; twelve of twelve in 2025/26 (minimum 5,316, Year 11). Forward cohorts read each 2025/26 year group as the Year 11 of the year it arrives there; cohorts tracked between the 2021 and 2026 DfE censuses held or grew through primary phase and eased 13–15 per cent on entering secondary. Totals: 32,143 → 72,017 (+124%) → 98,210 (+36%); free movement ended 31 December 2020, three weeks before the January 2021 collection, so the growth from the second to the third school and pupil census occurred entirely under the post-free-movement regime. ↩
      23. ONS commissioned table CT21_0503 (main language, 67 categories, by age, 331 lower-tier authorities), reconciling with TS024 and CT21_0418 to within disclosure-control perturbation; school-age = 3–14 at 21 March 2021. The instruments count different populations at different dates – pupils in January 2026 against resident speakers in March 2021 – so the ratio bundles the census undercount of the heritage cohort with five years of arrivals, not growth alone. ↩
      24. Secondary-phase (Years 7–11) pupils with Romanian as first language (FOI-2026-0093642) ÷ pupils with first language other than English, state-funded secondary schools, 2025/26 (DfE, Schools, pupils and their characteristics) × 100; computed for 151 authorities. An earlier iteration of this analysis, before the named-language counts were available, used a census-based intensity index (all-ages Romanian speakers per 100 secondary EAL pupils); the measured share supersedes it and confirms its tiering, with one promotion – Havering – into the upper tier. ↩
      25. School counts: DfE school-level underlying data, 2025/26. Birmingham 926 across 91 ≈ 10 per school; Harrow 1,093 across 13 ≈ 84; West Northamptonshire 1,066 across 23 ≈ 46; Thurrock 374 across 13 ≈ 29; North Northamptonshire 598 across 21 ≈ 28; Doncaster 296 across 20 ≈ 15. Even-spread division is a conservative floor for any individual school. ↩
      26. ONS, Census 2021, commissioned table (publicly available) CT21_0418 (main language, 67 categories, by proficiency in English by age; England and Wales): www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/populationestimates/adhocs/3005ct210418census2021. Author’s calculations: school-age counts sum the 3-9 and 10-14 bands of the local-authority table CT21_0503 (27,895 + 16,428 = 44,323; 9.4% of 471,954; the national table CT21_0418 gives 27,903 + 16,429 = 44,332 within disclosure-control perturbation); the indicative Year 11 cohort divides the 3-9 band across its seven birth years (≈3,985 per year), ageing forward from 21 March 2021; proficiency: 383,885 of 471,954 (81.3%) speak English well or very well. Comparator profiles computed identically. CT21_0418 is a national-level table; its local-authority equivalent, CT21_0503, commissioned for this report, is used throughout this edition. All figures subject to the caveats mentioned in the Data section, particularly the conservative measures of population count. ↩
      27. ONS commissioned table CT21_0503: main language (detailed, 67 categories) by age (17 bands), all usual residents aged 3 and over, 331 lower-tier local authorities, England and Wales; produced to this report’s specification. National figures are sums across authorities; small differences from previously published national tables reflect disclosure-control perturbation. School-age = 3–14 at 21 March 2021; Year 11 pipeline: the 3–9 band aged forward, ≈ 27,895 ÷ 7 birth-years ≈ 4,000 annually, 2027/28–2033/34. ↩
      28. Pupil counts: DfE, FOI-2026-0093642, section D. Home Office, Immigration system statistics, year ending June 2026, EU Settlement Scheme detailed tables (EUSS_D01, applications received; EUSS_D02, concluded applications by latest case outcome), published on the 27th of August 2026: www.gov.uk/government/statistical-data-sets/immigration-system-statistics-data-tables. Author’s extraction, nationality = Romania. Figures count applications, not unique individuals, and are cumulative since 2019; nationality is not a language measure, so the contextual comparison with census-school-age speakers is indicative/illustrative of certain measurement gaps, not a like-for-like count. ↩
      29. The figure is from the national table main language by English proficiency by age (ONS, CT21_0418): 383,885 of 471,954 speakers reporting English spoken well or very well. ↩

      Notes

      Illustrative standardisation, not a forecast. The implied entry range (Figure C1), the Year 11 entry check (section C), the cohort strip (Figure F2) and the pupil-to-census ratios (Figure D4) apply observed rates and ratios to Romanian counts to show scale. They do not predict take-up.

      The measured share supersedes the intensity index. Earlier editions ranked authorities by a census-based intensity index (all-ages Romanian speakers per 100 secondary pupils with a first language other than English). With the Department’s named-language counts, Figure E1 now shows the measured share: secondary pupils with Romanian as first language as a percentage of secondary EAL pupils, both from the January 2026 school census. The index stays selectable in Figure E1 for comparison; it gave the same tiering, with one promotion (Havering) into the upper tier.

      Census figures are point-in-time estimates for 21 March 2021 and are perturbed by ONS statistical disclosure control, so small counts may not sum exactly. All census-derived figures in this report are conservative: they count only residents who reported Romanian as their main language, and so exclude bilingual, UK-born children who report English. School counts are January headcounts in England’s state-funded schools; settlement-scheme figures count applications, not people.

      Download the full report (PDF) Descarcă raportul în limba română (PDF)

      Contains public sector information licensed under the Open Government Licence v3.0. Census data: Office for National Statistics. School data: Department for Education. Examination data: Joint Council for Qualifications and Ofqual. Boundaries: contains OS data © Crown copyright and database right 2021.

      ResearchAndra-Lucia Martinescu
      Data architectureMarius Dima

      Support the campaign

      We are run entirely by volunteers. A small donation helps us organise events, print materials and keep campaigning for a Romanian GCSE.

      Scroll to Top