U.S. PISA Results by Cycle: 2000–2025
Mean U.S. PISA scores for 15-year-old students across nine assessment cycles, with trend-comparable reading, mathematics, and science results and cycle-specific reporting limits.
U.S. PISA Scores Across Assessment Cycles
Trend-comparable national mean scores for reading, mathematics, and science.
Reading trend comparisons begin in 2000, mathematics in 2003, and science in 2006. U.S. reading is not reported for 2006 because of a test-booklet printing error; U.S. 2025 estimates carry an OECD data-quality flag.
OECD PISA 2025 trend tables; NCES PISA historical results.
Full U.S. PISA Results by Cycle
Scores are national means rounded to whole PISA score points. A dash marks a domain that is outside the comparable trend series for that cycle or, for U.S. reading in 2006, a result that was not reportable.
| Cycle Note | |||||
|---|---|---|---|---|---|
| 2000 | Reading | 504 | — | — | Reading trend baseline. |
| 2003 | Mathematics | 495 | 483 | — | Mathematics trend baseline. |
| 2006 | Science | — | 474 | 489 | Science trend baseline. U.S. reading was not reported because of a booklet-printing error. |
| 2009 | Reading | 500 | 487 | 502 | All three trend series reported. |
| 2012 | Mathematics | 498 | 481 | 497 | All three trend series reported. |
| 2015 | Science | 497 | 470 | 496 | U.S. administration moved fully to computer-based assessment. |
| 2018 | Reading | 505 | 478 | 502 | All three trend series reported. |
| 2022 | Mathematics | 504 | 465 | 499 | The cycle originally planned for 2021 was delayed to 2022. |
| 2025 | Science | 490 | 463 | 502 | OECD flags U.S. estimates because one or more sampling standards were not met. |
Reading Fell in 2025 While Science Remained Stable
- Reading stayed near 500 points from 2000 through 2022, then moved to 490 in 2025. OECD reports the 2025 reading result as statistically lower than 2022, while the latest mathematics and science means were not statistically distinguishable from 2022.
- Mathematics shows a longer downward direction from its 2003 trend baseline of 483 to 463 in 2025. Science has been much flatter since its 2006 baseline, remaining between 489 and 502 across the reported cycles.
- The three lines do not share the same starting year. Comparable reading trends begin in 2000, mathematics in 2003, and science in 2006; the missing U.S. reading point in 2006 is a reporting break rather than a zero or estimated score.
- Extra caution applies to 2025. U.S. school and student participation, together with the exclusion rate, did not meet all PISA technical standards, and the available analyses cannot determine the direction of any resulting bias.
Data Sources, Measurement and Comparability
Official Results
The cycle means and 2025 comparison statements use OECD and NCES PISA reporting. Primary references include the OECD PISA 2025 United States country note, OECD Education GPS for the United States, and the OECD PISA 2022 U.S. trend overview.
Trend Baselines
Cross-cycle comparisons begin when each domain first served as a PISA major domain on the current trend scale: reading in 2000, mathematics in 2003, and science in 2006. Earlier minor-domain results are not inserted into these trend lines. The NCES PISA 2025 Methodology and Technical Notes also documents the linking errors used for comparisons with earlier cycles.
2006 U.S. Reading
The United States has no reportable PISA 2006 reading mean. A booklet-printing error caused incorrect page-reference instructions in parts of the reading assessment, and the affected score could not be recalibrated. NCES documents the issue in PISA 2006 U.S. results and the related technical reporting.
Rounding and Statistical Testing
Displayed means are rounded to whole score points. Cross-cycle tests use the underlying unrounded estimates together with sampling error and PISA linking error, so subtracting two displayed whole-number scores does not always reproduce the officially reported change. A numerical difference should not be treated as statistically distinguishable unless the OECD or NCES comparison supports that interpretation.