The Netherlands entered PISA in 2000, but its usable trend series starts later because the 2000 sample did not meet OECD response-rate standards. By PISA 2025, mathematics remained above the OECD average, science was close to it, and reading was below it. The longer record is more striking: all three core domains are well below their earlier Dutch levels, with the deepest decline in reading.
PISA 2025 places the Dutch subjects in three different positions
Mean scores for 15-year-old students, with the matching OECD average and OECD-published change from PISA 2022.
The 2025 pattern is uneven. OECD classifies Dutch mathematics performance as above the OECD average, science as close to the OECD average, and reading as below it. For the latest cycle change, mathematics and science were not statistically distinguishable from their 2022 results, while reading was lower in 2025. OECD also reports that all three Dutch core-domain results were lower than in 2018 and that the direction since 2015 is downward.
Netherlands and OECD average in PISA 2025
The gap column is calculated from the published rounded mean scores. The cycle-change column uses OECD’s published change measure.
Scroll horizontally to view all columns.
| Domain | Netherlands | OECD average | Arithmetic gap | 2022–2025 OECD change | OECD comparison |
|---|---|---|---|---|---|
| Mathematics | 483 | 463 | +20 | −9 | Above OECD average |
| Science | 485 | 482 | +3 | −3 | Close to OECD average |
| Reading | 441 | 461 | −20 | −18 | Below OECD average |
Being above the OECD average does not mean the Netherlands has returned to its own earlier performance. Mathematics is 20 points above the 2025 OECD average, yet its published Dutch mean is 55 points below the 2003 level.
Netherlands PISA scores by cycle, 2000–2025
The table uses the comparable historical series for each domain rather than filling non-comparable years with nominal scores.
Historical Dutch mean scores
Reading and mathematics are shown from 2003 for the Netherlands; the comparable science series starts in 2006.
Scroll horizontally to view all columns.
| PISA cycle | Mathematics | Reading | Science | Trend status |
|---|---|---|---|---|
| 2000 | — | — | — | No comparable Dutch national mean used for trend analysis |
| 2003 | 538 | 513 | — | Mathematics and reading series available |
| 2006 | 531 | 507 | 525 | First comparable science point |
| 2009 | 526 | 508 | 522 | Comparable cycle |
| 2012 | 523 | 511 | 522 | Comparable cycle |
| 2015 | 512 | 503 | 509 | Comparable cycle |
| 2018 | 519 | 485 | 503 | Comparable cycle; OECD annotation applies to Dutch sampling |
| 2022 | 493 | 459 | 488 | Comparable cycle; OECD sampling caution applies |
| 2025 | 483 | 441 | 485 | Latest cycle; OECD sampling caution applies |
A dash means a comparable trend value is not used for that domain and cycle. It does not mean a score of zero.
The long-run decline is deepest in reading
Comparable Dutch mean scores across PISA cycles
The vertical scale runs from 400 to 560 to keep cycle differences readable without implying that zero is a meaningful PISA baseline. Science starts in 2006 because that is the first comparable point in the science trend series.
- Reading was broadly stable around 507–513 from 2003 through 2012, then fell in every reported cycle from 2012 to 2025: 511, 503, 485, 459 and 441.
- Mathematics did not fall in a straight line. It rose from 512 in 2015 to 519 in 2018 before dropping to 493 in 2022 and 483 in 2025.
- Science moved from 525 in 2006 to 485 in 2025, but the latest 2022–2025 movement was only −3 points in OECD’s published change measure.
Why the Dutch trend cannot start with a 2000 score
The Netherlands did participate in PISA 2000. The problem is comparability, not participation. OECD’s PISA 2000 technical documentation reports a weighted school participation rate of 27.13% before replacement for the Netherlands, far below the required standard. OECD therefore did not report a Dutch national mean for use in PISA 2000 trend comparisons.
For the Netherlands, the usable historical series begins in 2003 for mathematics and reading and in 2006 for science. The page range can still begin in 2000 because that is when the Netherlands first participated, but a numeric 2000 point should not be inserted into the trend line.
This distinction matters because a trend chart that assigns a Dutch reading, mathematics or science mean to 2000 would create a false baseline. It would also make the first-to-latest change look more precise than the official record allows.
The break in the Dutch pattern becomes clear after 2012
2003–2012: high scores with limited movement in reading and science
Mathematics declined from 538 in 2003 to 523 in 2012. Reading moved from 513 to 511 over the same endpoints, while science was 525 in 2006 and 522 in 2012. The early period therefore does not support a simple story of uninterrupted collapse across all subjects.
2012–2018: reading moves away from the earlier range
Reading fell from 511 in 2012 to 503 in 2015 and 485 in 2018. Science also moved down, from 522 to 509 to 503. Mathematics followed a different path: 523 in 2012, 512 in 2015 and then 519 in 2018. That 2018 mathematics rebound is one reason the three Dutch subject histories should not be compressed into a single overall PISA score.
2018–2022: all three domains move lower
The OECD-published 2018–2022 changes were −26.6 points in mathematics, −25.5 in reading and −15.1 in science. The 2022 means were 493, 459 and 488 respectively. This cycle shifted the Dutch profile downward across all three core domains at the same time.
2022–2025: reading falls again while math and science are broadly stable
PISA 2025 did not produce a broad Dutch rebound. OECD reports changes of −9 in mathematics, −18 in reading and −3 in science from 2022. The mathematics and science changes are not statistically distinguishable from zero in OECD’s comparison, while reading is lower than in 2022. The latest cycle therefore extends the reading decline more clearly than the other two domain trends.
Reading has lost the most ground
First comparable score to PISA 2025, calculated from OECD-published rounded means.
Reading is the clearest long-term weakness in the Dutch series. The 72-point decline from the first comparable 2003 mean to 2025 is larger than the corresponding mathematics decline, and the reading series has dropped in every cycle since 2012. From 2012 alone, the published mean moved from 511 to 441, a 70-point decline.
Mathematics presents a different relative picture. Its long-run Dutch decline is large, yet the 2025 mean of 483 still sits 20 points above the OECD average of 463. Science has also moved well below its mid-2000s level, but its 2025 mean of 485 remains close to the OECD average of 482.
The score decline is also visible in proficiency levels
PISA 2025 shares at or above Level 2 and at the top proficiency levels.
PISA 2025 proficiency profile
Level 2 is OECD’s baseline proficiency reference. Top performers are Levels 5–6; for reading, OECD reports Level 5 or higher.
Scroll horizontally to view all columns.
| Domain | Netherlands Level 2+ | OECD Level 2+ | Netherlands top performers | OECD top performers |
|---|---|---|---|---|
| Science | 73% | 74% | 9% | 7% |
| Mathematics | 70% | 65% | 13% | 8% |
| Reading | 61% | 69% | 4% | 6% |
Low-performing shares rose most in reading from 2015 to 2025
Increase in the share of Dutch students below Level 2, in percentage points
OECD reports the change in the below-Level-2 share from PISA 2015 to PISA 2025. Percentage points measure a change in a share, not a percent change.
The reading problem is visible in both the mean and the distribution. Only 61% of Dutch students reached Level 2 or higher in reading in 2025, compared with 69% across OECD countries, while the share below Level 2 had risen by 20 percentage points since 2015.
These two cross-domain shares show why a single national mean cannot describe the whole Dutch performance distribution. The Netherlands had a larger share of students reaching Level 5 or 6 in at least one core domain than the OECD average, while it also had a somewhat larger share below Level 2 in all three domains.
Why 493 to 483 is not the official −10 trend figure
The displayed mathematics means are 493 for 2022 and 483 for 2025, so subtracting the rounded values gives −10. OECD’s published short-term change is −9 points. This is not a contradiction: OECD calculates totals, differences and averages from exact underlying estimates and rounds only after the calculation. For trend reporting, the OECD-published change should therefore take precedence over subtraction of the displayed rounded means.
Displayed score arithmetic is useful for simple descriptive comparisons. When OECD publishes a dedicated trend difference, that value is the better measure because it is calculated from the unrounded estimates.
PISA 2025 adds computational problem solving, but it is not part of the historical core trend
The Netherlands scored 506 in computational problem solving in PISA 2025, compared with an OECD average of 500. OECD reports the Dutch result as above the OECD average. This measure belongs to PISA 2025’s Learning in the Digital World assessment and should not be inserted into the long-run mathematics, reading or science chart as if it were another point in those established scales.
The 2025 Dutch estimates carry an OECD sampling caution
The Netherlands is marked with an asterisk in PISA 2025 because one or more sampling standards were not met. The overall exclusion rate was 9.3%. The weighted school response rate was 66% before replacement and 88% after replacement, while the student response rate was 78%.
OECD’s range-of-bias analysis indicates that student non-response could affect the Dutch science mean by about 11 score points in either direction under hypothetical best- and worst-case scenarios, even after weighting adjustments. At school level, OECD reports much smaller potential bias after replacement schools and adjusted weights. These figures do not make the Dutch results unusable, but they do mean that small differences should be read with more caution than the rounded score alone suggests.
The Netherlands remains in OECD’s published PISA 2025 tables. The asterisk is a data-quality annotation, not a deletion of the result. It signals extra uncertainty because sampling standards on exclusions and participation were not fully met.
What the 2000–2025 record says about Dutch performance
The Netherlands still has a relative strength in mathematics: its 2025 mean is above the OECD average and its share of mathematics top performers is also higher than the OECD average. That relative position coexists with a large decline from the Dutch 2003 mathematics mean.
Reading is the clearest long-run deterioration. The comparable series falls 72 points from 2003 to 2025, and the 2025 reading result is lower than both the Dutch 2022 result and the 2025 OECD average. The rise in students below Level 2 since 2015 points in the same direction.
Science sits between those two stories. Its 2025 mean is close to the OECD average and changed little from 2022, yet it remains 40 points below the first comparable Dutch science score in 2006. Taken together, the historical record is best described as a broad retreat from earlier Dutch performance, with the largest and most persistent loss in reading.
Data Sources and Measurement
Primary OECD material for the Netherlands’ 2025 comparison with the OECD average, proficiency shares, trend interpretation, sample description and country-specific findings. OECD PISA 2025 Results: Netherlands
Used for 2025 mean scores, OECD averages, official 2022–2025 score changes, computational problem solving, and cross-domain top- and low-performer shares. OECD PISA 2025 Volume I snapshot tables
PISA 2003–2022 Dutch mathematics, reading and science means are based on OECD’s performance-trend tables; 2025 is added from the latest OECD release. OECD PISA performance trends
OECD documentation explains why the Netherlands is excluded from PISA 2000 trend comparisons and why published differences can differ from subtraction of rounded means. OECD PISA FAQs · PISA 2025 Reader’s Guide
The PISA 2000 technical report records the Netherlands’ weighted school participation rate before replacement as 27.13%, explaining why the 2000 result is not used as a Dutch trend baseline. OECD PISA 2000 Technical Report
PISA mean scores are population estimates for 15-year-old students in the PISA target population, not pass/fail marks or an individual student’s proficiency level. Historical changes calculated directly from displayed means are descriptive arithmetic; OECD-published trend differences are preferred when available because OECD calculates differences from exact estimates before rounding.