Primary schools

Which schools and pupils?

The directory uses the DfE school-level underlying data for January 2003, 2005, 2010, 2015, 2016, 2025 and 2026, identifying the primary phase from each release’s published school-type and phase fields. Historical releases identify maintained primary schools; later releases use the “State-funded primary” classification, including primary academies. Independent schools, special schools and institutions classified in another phase are outside this directory. Primary-age pupils in all-through schools classified as secondary cannot be separated using these whole-school ethnicity counts.

Comparing the snapshots

Published ethnicity returns for 2003, 2005, 2010, 2015 and 2016 cover pupils of compulsory school age and above. The 2025 and 2026 returns cover all ages, including younger pupils. Dated shares remain visible, but percentage-point changes across this coverage boundary are withheld. Within the same coverage basis, a comparison requires the same school URN or a unique, unmerged official predecessor chain. New schools, ambiguous histories and missing records are not treated as zero. These are different pupil populations, not the same children followed through time.

Counts, percentages and missing values

Displayed percentages divide the published group count by all pupils in the relevant census population. Unclassified pupils and any unallocated gap in the source category counts remain in the denominator. They are never added to White British or assigned to another ethnicity. Unclassified remains a count in school records and downloads rather than a percentage filter. The 2003 and 2005 files publish the compulsory-school-age population directly. For 2010, exact populations are verified from the available age fields; rounded published headcounts are not silently treated as exact. Where protected age cells prevent verification, a school’s separately published percentage is labelled as such. The 2015 and 2016 populations are verified from DfE headcounts and age fields. Available original source percentages are checked against the appropriate population. Protected values (# or ~ in the historical files) remain unavailable. A pooled share is withheld when any contributing selected count or denominator is missing; a partial sum is not presented as the full area or school population. Where protected school-level cells prevent a complete area total, a separately published DfE total for the same phase and population may supply the England or local-authority share. The 2003 and 2005 primary area totals use the official maintained-primary tables for that source population. These figures are only used for the full source population, never a school search. School records retain the archived school-level phase codes. Separately published aggregate totals follow the publication’s phase allocation, which differs for two schools in 2005, so those totals need not equal the sum of the school rows shown here. The ≈ symbol marks a source population rounded by the publisher; changes involving these rounded aggregates are withheld. Area totals may be available even when individual school counts are protected. Where the publisher gives a school percentage but protected age cells prevent verification of an exact population denominator, that published percentage is shown with its source label. It is not an inferred count and is not used in a pooled share or percentage-point change. The download keeps original counts, DfE source percentages, age coverage, share method, denominator and quality fields separately.

Historical rows with a missing ethnicity breakdown are flagged and withheld, even if their source cells contain zero. Their raw source cells remain in the download. A genuine recorded zero for one ethnic group remains 0% when the census population is positive and a classified ethnicity breakdown is reported for other groups. If no classified breakdown is available, including returns recorded entirely as Unclassified, the ethnicity share is withheld rather than presented as 0%. No ethnicity is inferred for unknown pupils. Zero or unavailable population denominators, protected counts and missing records remain unavailable. The ethnic-group percentages need not add to 100% because unknown pupils remain in the population.

A different question from end of primary

This page describes whole-school January census returns at primary institutions. The separate end-of-primary map describes local-authority KS2 assessment cohorts, usually Year 6. Neither series supplies the missing Year 7-by-ethnicity breakdown.

Original data

• 2003 · DfE January 2003 school-level census ↗
• 2005 · DfE January 2005 school-level census ↗
• 2010 · DfE January 2010 school-level census ↗
• 2015 · Underlying data: SFR16/2015 ↗
• 2016 · Underlying data: SFR20/2016 ↗
• 2025 · Download School level underlying data 2025 (csv, 22 Mb) ↗
• 2026 · Download School level underlying data 2026 (csv, 22 Mb) ↗

Data catalogue and method notes (JSON) ↓ · Get Information about Schools ↗

Department for Education data, reused under the Open Government Licence v3.0.