A KoinEd Update

Koin-Ed takes a three-year read of the Department for Education’s provisional key stage 2 attainment statistics for 2025/26, published 10 September 2026.

The provisional key stage 2 figures for summer 2026 have landed and attainment in reading, writing and maths combined rose again. Nothing fell back badly. The disadvantage gap widened. Read on its own, the release tells a modest good-news story with one persistent drawback.

A single year of comparison is a poor instrument for a system recovering from an interruption as large as the pandemic. Set 2026 against 2025 and 2024 together and a more useful picture appears: consistent, small gains at the expected standard, almost no movement at the top end, and a distribution that is pulling apart at the same time as the average creeps up.

This post works through the three cohorts side by side, separating what is really improving from what is drifting in the wrong direction, and flags the places where the headline number is doing more work than it should.

Why 2024, 2025 and 2026 belong together

These three cohorts sat the first fully re-established set of assessments after the pandemic. All three were in primary school during the closures, and the 2026 cohort is described by the DfE as having lost most learning time at the end of year 1 and during year 2. They are therefore comparable to one another in a way that none of them is comparable to 2019.

One caution before the numbers. The three years sit at different stages of the statistical process: 2024 is final data, 2025 is revised, and 2026 is provisional and will be updated in December 2026. Small movements of a few tenths of a percentage point should not be pressed hard. The DfE also calculates all gaps and changes from unrounded figures, so a rise from 72% to 73% may be anything from 0.5 to 1.4 percentage points in reality.

The headline picture

At the expected standard, every measure is either level or up across the three years.

Table 1: Pupils meeting the expected standard, England, 2024 to 2026

Sources: DfE key stage 2 attainment, provisional 2025/26 and 2024/25 releases. The 2019 column is shown only where the DfE treats the series as directly comparable.

Two years of movement have added two percentage points to the combined measure and to maths, one to reading, writing and science, and two to grammar, punctuation and spelling. Reading remains the only subject that has both recovered its pre-pandemic level and gone beyond it.

The positive trends

1. A third consecutive year of improvement in the headline measure

Sixty-one per cent in 2024, 62% in 2025, 63% in 2026. The direction has held for three years running, and the 2025 rise of roughly 1.6 percentage points was the largest recorded since assessments resumed. That is a slow climb, but it is a climb, and it has not stalled or reversed at any point in the window.

2. Maths is the clearest recovery story

Maths has improved on every measure in every year of the three. The expected standard went 73%, 74%, 75%. The higher standard went 24%, 26%, 27%. The average scaled score rose from 104 to 105 and held there. No other subject shows that consistency, and maths is the only subject where the top end has moved at the same time as the threshold.

3. Writing has finally started to move

Writing was flat between 2023 and 2024 and rose only about half a percentage point in 2025, not enough to change the rounded figure. In 2026 it reached 73%. This matters more than the size of the increase suggests, because writing has been the subject preventing most pupils from achieving the combined measure since 2022.

4. Grammar, punctuation and spelling has climbed steadily

Seventy-two per cent, 73%, 74% at the expected standard. Two percentage points across the window, and the most reliable year-on-year pattern of any of the tested subjects apart from maths.

5. The higher standard in the combined measure has finally shifted

The proportion reaching the higher standard in reading, writing and maths together sat at 8% in 2023, 2024 and 2025. In 2026 it reached 9%. Three years of no movement followed by a rise is worth noting, even if the absolute level remains very low.

6. Fewer pupils are failing the assessments they sit

The proportion assessed as not meeting the expected standard in the combined measure fell from 31.1% in 2024 to 29.3% in 2025 and 28.0% in 2026. In headcount, 188,741 pupils were assessed as not meeting the standard in 2025 against 181,771 in 2026, a fall of nearly 7,000 despite broadly stable cohort sizes. The proportion assessed as failing all three subjects also fell, from 8.5% to 7.9%.

7. Pupils with special educational needs improved on every measure

This is the strongest sustained improvement anywhere in the release.

Table 2: Pupils meeting the expected standard in reading, writing and maths (combined), by SEN status

Four percentage points on the SEN group and five on SEN support in two years, against two percentage points for the cohort as a whole. Pupils with SEN are improving faster than the average, and they are doing so while the SEN share of the cohort grows, which normally works against the figure.

8. Pupils with English as an additional language are pulling ahead

Pupils whose first language is not English went from 62% to 64% to 66% on the combined measure, while pupils with English as a first language went from 60% to 62% to 63%. In maths the gap now runs five percentage points in favour of pupils with EAL, and in writing they are two points ahead. The long-standing reading deficit has almost closed, narrowing to around one percentage point in 2026.

9. The summer-born penalty narrowed slightly

In 2025, September-born pupils reached 69% on the combined measure and August-born pupils 55%, a spread of 14 percentage points. In 2026 the figures were 70% and 57%, a spread of 13. August-born pupils gained two points against the cohort’s one. One year does not make a trend, but the movement is in the right direction on a gap that has proved stubborn for a decade.

10. Regional and local authority floors have risen

London remains well clear of everywhere else, at 70% in 2026 against 68% in 2025. The more useful observation is what happened underneath it. In 2025 the other regions ranged from 59% to 61%; in 2026 they ranged from 61% to 63%. The whole distribution shifted up by roughly two points rather than London pulling further away. The gap between the strongest and weakest regions was ten percentage points in 2024 and nine in both 2025 and 2026.

At local authority level the lowest-attaining authority moved from 50% in 2025 to 53% in 2026, and the highest from 76% to 78%. The floor rose faster than the ceiling.

The negative trends

1. The below-curriculum group keeps growing, and it flatters the headline

This is the single most important qualification on the whole release. A growing number of pupils are deemed to be working below the requirements of the key stage 2 curriculum in at least one subject. They do not sit the relevant test or receive the standard teacher assessment, so they cannot be recorded as having failed it.

That proportion was 5.1% before the pandemic in 2019. It was 7.8% in 2025 and 8.1% in 2026, and it has risen every single year. Add pupils who were absent or otherwise without a result and 8.7% of the 2026 cohort were unable to meet the expected standard for reasons other than being assessed and falling short, up from 8.4% the year before.

The DfE says so itself: while the proportion meeting the expected standard has risen steadily since assessments returned, so has the proportion working below the requirements of the curriculum, and this may partly explain why attainment remains below pre-pandemic levels.

In headcount terms, 21,305 pupils were below the requirements in all three subjects in 2025 and 22,434 in 2026. Pupils below in writing alone went from 23,933 to 25,196. So of the movement in the combined measure between 2025 and 2026, a little over a percentage point is genuine improvement and roughly 0.3 percentage points reflects more pupils moving outside the assessed population altogether.

The composition of that group is also narrowing. In 2025, 81% of pupils working below the requirements in at least one subject had a special educational need. In 2026 it was 84%.

2. The disadvantage gap index has widened, and the 2026 jump is the largest of the three years

Table 3: Disadvantage gap index

A gap index of zero would mean no difference in average performance between disadvantaged pupils and their peers. Higher means a wider gap. The 2026 figure is provisional.

The index moved 0.03 between 2024 and 2025 and 0.06 between 2025 and 2026. Whatever has driven the improvement in average attainment has not reached disadvantaged pupils at the same rate. The gap is now substantially wider than it was in 2019, and wider than at any point since 2012 apart from the immediate post-pandemic years.

The provisional figure understates the position. It excludes children in local authority care unless they were eligible for free school meals in the last six years or ceased to be looked after in the last year. The DfE expects the revised December figure to be around 0.02 higher, which would put the real 2026 gap near 3.24.

Two supporting details make the picture worse. The attainment gap at the expected standard held at 22 percentage points for the second year running, so a rising tide is not narrowing it. And in reading, the one subject where the system has beaten its pre-pandemic level, attainment for disadvantaged pupils actually fell by around half a percentage point, widening the reading gap from 17 to 18 points.

Underneath that, the share of disadvantaged pupils working below the requirements of the curriculum rose from 13.3% to 13.9%, while the figure for other pupils stayed flat at 5.3%. The entire growth in the below-curriculum group sits on the disadvantaged side of the ledger.

3. Reading has stalled, and the top end went backwards

Reading held at 75% at the expected standard and 106 for the average scaled score. At the higher standard it went 29% in 2024, 33% in 2025, then back to 31% in 2026. The 2025 figure was the highest since the current curriculum assessments began in 2016, and roughly half of that gain has now been given back. The boys-to-girls gap at the higher standard in reading moved with it, narrowing from seven points to five and then widening again to six.

4. Writing at greater depth has not moved for five years

Thirteen per cent in 2022, 2023, 2024, 2025 and 2026. In both 2018 and 2019 the figure was 20%. Writing remains the weakest of the three core subjects at both the expected and the higher standard, and this is the clearest example in the release of a measure that has simply stopped responding.

5. Grammar, punctuation and spelling has lost ground at the top

The higher standard went 32% in 2024, 30% in 2025, 30% in 2026, so the expected standard has risen two points over the window while the higher standard has fallen two. The average scaled score has been 105 every year since 2022.

6. The maths gap between boys and girls is widening at the higher standard

Five percentage points in 2024, eight in 2025, nine in 2026. In 2026, 32% of boys and 22% of girls met the higher standard in maths. Boys have gained two points in a year; girls have gained none. This is the fastest-moving gap in the release and it has moved in the same direction for two years.

Elsewhere the picture by sex is more settled. Girls outperform boys in every subject except maths. The widest gap at the expected standard remains writing, though it narrowed from 12 points to 11. On the combined measure girls went from 65% to 67% and boys from 59% to 60%, holding the gap at six points, which means boys gained less than girls in a year when the headline improved.

7. Threshold measures are rising while typical performance is not

Table 4: Average scaled scores

Scaled scores rose in 2025 and then held. The expected standard rose again in 2026, but the average score of a typical pupil did not. Scaled scores are a blunt instrument at this level of rounding, so this is a signal rather than a finding, but it is consistent with the gains being concentrated among pupils sitting close to the 100 threshold rather than reflecting a general lift across the ability range.

8. Recovery is still incomplete almost everywhere

The combined measure is 63% against 65% in 2019. Writing at greater depth is seven points down on 2019. The disadvantage gap index is 0.31 higher than 2019. The share working below the requirements of the curriculum is three points higher. Reading is the only subject that has cleared its pre-pandemic mark, at 75% against 73%. Seven years on from 2019, one indicator out of many has fully recovered.

9. Writing has replaced reading as the blocker

Before the pandemic, reading was the subject most often stopping pupils from achieving the combined measure: 7% of those assessed in all three subjects missed out on reading alone. In 2019, only 3% of pupils met the standard in reading and maths but not writing. By 2026 that figure is around 5%, while the share meeting the standard in writing and maths but not reading has fallen to about 4%. The bottleneck has moved, and it has moved to the subject with the least reliable measurement and the least movement at the top end.

Three things that look like trends but are not

School type

Local authority maintained schools fell from 54% of state-funded mainstream primaries in 2025 to 51% in 2026, while converter academies rose from 33% to 36% and free schools grew from 237 to 257. Because schools move between categories, year-on-year changes in attainment by school type capture the churn as much as the performance. Free schools remain the highest attaining group, but at 2% of schools their figures will bounce around. Academies as a group and local authority-maintained schools remain broadly similar to one another, as they were in 2025.

Which region or authority comes last

The lowest-performing region was the South West in 2025 and the East Midlands in 2026, but every region outside London sat within a two-point band in each year. The same applies to local authority extremes, which shuffle annually. Portsmouth appearing at the bottom for the higher standard in both years, at 3%, is more informative than any single-year ranking.

Ethnicity comparisons across the two releases

Chinese pupils remain the highest attaining group, rising from 79% to 82%, with Indian pupils second at 76%. At the other end, the 2025 release identified Gypsy and Irish Traveller pupils together at 20% while the 2026 release names Gypsy pupils at 24%. The grouping in the commentary differs, so this is not a clean year-on-year comparison and should not be read as a four-point improvement.

What to hold lightly

• The 2026 figures are provisional. Revised statistics are due in December 2026, and the disadvantage figures in particular will change once children in local authority care are fully included.
• Writing and the combined measure are not comparable with 2016 and 2017 because the teacher assessment frameworks changed in 2018. Science is not comparable with years before 2019 for the same reason.
• There are no 2020 or 2021 figures, so every ‘since the pandemic’ comparison is really a comparison with 2022.
• Writing and science rest on teacher assessment rather than a test, which makes them more sensitive to changes in moderation practice than the tested subjects.
• National, regional and local authority figures are produced on slightly different methodologies, so pupil and school counts will not always reconcile.

What December will tell us

The revised release in December 2026 carries more weight than usual for two reasons. School-level data arrives, which allows the national averages to be interrogated properly. More importantly, key stage 1 to key stage 2 progress measures return for the first time since 2022/23. They were suspended for the 2023/24 and 2024/25 cohorts because those pupils had no key stage 1 assessments. The 2025/26 cohort does.

That matters because every trend described above is an attainment trend, and attainment measures where pupils finished without saying anything about where they started. Progress data will show whether the combined measure’s climb reflects schools adding more value, or a cohort arriving at year 6 in slightly better shape than the two before it. It will also give the first real test of whether the growth in the below-curriculum group is a change in pupil need or a change in assessment practice.

The three-year summary

Attainment is rising slowly and consistently. Maths and grammar, punctuation and spelling are the engines; writing has begun to move; reading has plateaued after a strong 2025 and slipped at the top. Pupils with SEN and pupils with English as an additional language are improving faster than the cohort as a whole.

Against that, the proportion of pupils sitting outside the standard assessment has risen every year since 2019 and is now large enough to influence how the headline should be read. The disadvantage gap widened more in 2026 than in the previous two years combined, and the growth in the below-curriculum group falls almost entirely on disadvantaged pupils. The top end of the distribution has barely moved in five years, and the maths gap between boys and girls at the higher standard is widening quickly.

A system recovering unevenly is still a system recovering. But the two numbers to watch over the next three years are not the headline percentage. They are the share of pupils working below the requirements of the curriculum, and the disadvantage gap index.

Simin Uttley

Director,

Koin-Ed.org

Source: Department for Education, Key stage 2 attainment, Academic year 2025/26 (provisional), published 10 September 2026, and Key stage 2 attainment, Academic year 2024/25, published 11 September 2025. Contains public sector information licensed under the Open Government Licence v3.0.