PISA proficiency levels show how many 15-year-old students can perform tasks at different levels of difficulty, rather than reducing performance to a single mean score. In PISA 2025, Level 2 is the baseline proficiency benchmark in science, mathematics and reading, while students at Levels 5 and 6 are classified as top performers.
PISA 2025 Proficiency Levels by Subject
OECD-average shares of students reaching baseline and advanced proficiency in the three core PISA subjects.
Baseline attainment was highest in science among OECD countries in 2025: 74% reached at least Level 2, compared with 69% in reading and 65% in mathematics. The pattern at the upper end was different. Mathematics had the largest OECD-average share at Levels 5–6, at 8%, compared with 7% in science and 6% in reading.
OECD-Average Proficiency Distribution
The chart separates students below the Level 2 baseline, students at Levels 2–4, and students at Levels 5–6.
Levels 2–4 are calculated from the OECD-published rounded shares for students below Level 2 and at Levels 5–6. Percentages may therefore be affected by rounding.
Mathematics had both the largest low-performer share and the largest top-performer share among the three subjects. That combination illustrates why a mean score alone does not describe the shape of a performance distribution: the lower and upper ends can both differ materially across subjects.
PISA 2025 Proficiency Levels by Country and Subject
Selected cross-country profiles using OECD-published PISA 2025 country-note percentages. Level 2+ is the share reaching at least baseline proficiency; Levels 5–6 identifies top performers.
| Country / economy | Science | Mathematics | Reading | |||
|---|---|---|---|---|---|---|
| Level 2+ | Levels 5–6 | Level 2+ | Levels 5–6 | Level 2+ | Levels 5–6 | |
| Australia | 82% | 12% | 71% | 10% | 77% | 11% |
| B-S-J-Z (China) | 97% | 36% | 96% | 54% | 93% | 10% |
| Canada* | 84% | 11% | 74% | 11% | 78% | 9% |
| Chinese Taipei | 87% | 20% | 85% | 32% | 82% | 13% |
| Estonia | 90% | 11% | 83% | 13% | 83% | 9% |
| Finland | 81% | 11% | 69% | 7% | 74% | 6% |
| Germany | 73% | 9% | 66% | 8% | 68% | 7% |
| Japan | 89% | 17% | 84% | 22% | 81% | 11% |
| Korea | 86% | 15% | 81% | 23% | 82% | 10% |
| Macao (China) | 92% | 14% | 90% | 29% | 85% | 7% |
| Singapore | 90% | 26% | 89% | 37% | 86% | 22% |
| Türkiye | 81% | 7% | 64% | 8% | 76% | 4% |
| United Kingdom | 83% | 13% | 74% | 13% | 80% | 9% |
| United States* | 78% | 13% | 65% | 8% | 74% | 13% |
| OECD average | 74% | 7% | 65% | 8% | 69% | 6% |
Percentages are rounded as reported in OECD PISA 2025 country notes. Levels 5–6 are included within the Level 2+ percentage, so the two columns for a subject should not be added together. An asterisk marks an education system for which the OECD advises caution because one or more PISA sampling standards were not met. OECD Annex Tables I.B1.2a.10, I.B1.2a.11 and I.B1.2a.12 are the authoritative sources for the complete participating-country/economy distributions at every proficiency level.
The Global Range in Baseline Proficiency
Country differences are much larger than the differences among the three OECD averages.
- B-S-J-Z (China) combined very high baseline attainment with an unusually large upper tail in mathematics: 96% reached Level 2 or higher and 54% reached Levels 5–6.
- Singapore also combined broad baseline attainment with large top-performer shares, particularly in mathematics, where 89% reached Level 2+ and 37% reached Levels 5–6.
- A high Level 2+ share does not imply an equally large Level 5–6 share. Macao (China), for example, had 92% at Level 2+ in science but 14% at Levels 5–6, while Singapore had 90% at Level 2+ and 26% at Levels 5–6.
Subject Profiles Can Differ Within the Same Education System
The three proficiency scales measure different domains, so the most useful comparison is the share of students reaching the corresponding proficiency benchmark within each subject.
Several systems show relatively similar baseline-attainment rates across science, mathematics and reading. B-S-J-Z (China), for example, recorded Level 2+ shares of 97%, 96% and 93%, while Singapore recorded 90%, 89% and 86%.
Other systems have a wider subject spread. In Türkiye, 81% reached Level 2 or higher in science and 76% did so in reading, compared with 64% in mathematics. In the United States, the corresponding shares were 78% in science, 74% in reading and 65% in mathematics. These are descriptive within-system contrasts; the subject scales should not be treated as if one PISA score point or proficiency threshold has identical substantive meaning in all three domains.
Upper-tail profiles can differ even more than baseline attainment. The United States had an 8% mathematics top-performer share but 13% in both science and reading. Macao (China), by contrast, had 29% at Levels 5–6 in mathematics, compared with 14% in science and 7% in reading.
What Level 2 Means in Science, Mathematics and Reading
Level 2 is the baseline used by the OECD to distinguish students who demonstrate foundational proficiency from students classified as low performers.
Students can use basic scientific knowledge in familiar situations, recognize plausible explanations, work with straightforward evidence and judge simple conclusions from the information provided. PISA 2025 treats this as the point at which students begin to use science rather than only recall scientific information.
Students can recognize how a simple real-world situation can be represented mathematically without being given direct instructions. OECD examples include comparing distances between alternative routes and converting prices between currencies.
Students can identify the main idea in a text of moderate length, locate information using stated criteria and reflect on a text’s purpose or form when prompted to do so.
Low Performers and Top Performers
A student scoring below Level 2 is classified as a low performer for that subject. The low-performer rate is therefore the complement of the Level 2+ rate. At the OECD average, 26% of students were below Level 2 in science, 35% in mathematics and 31% in reading in PISA 2025.
Top performers are students who reach Levels 5 or 6. Their share captures a different part of the distribution. A system can have relatively few students below Level 2 without having an exceptionally large share at Levels 5–6, while another system can combine both broad baseline attainment and a large upper-performing group.
Do not read Level 2+ and Levels 5–6 as separate populations. A student at Level 5 or 6 is also above Level 2. For example, Singapore’s 89% mathematics Level 2+ rate already contains the 37% of students classified as mathematics top performers.
Why Proficiency Shares Add Information Beyond Mean Scores
Mean PISA scores describe the center of a country’s performance distribution. Proficiency shares describe how students are distributed around policy-relevant skill thresholds. Two systems with similar mean scores can therefore have different percentages below Level 2 or at Levels 5–6.
This distinction matters when comparing education systems. A change in the mean does not by itself show whether movement occurred mainly among lower-performing students, students near the middle of the distribution or students at the upper end. PISA 2025 publishes both mean performance and proficiency distributions for this reason.
Coverage and Comparability Notes
- PISA 2025 included 91 participating countries and economies and more than 760,000 students, representing about 33 million 15-year-olds.
- Science was the major assessment domain in PISA 2025; mathematics and reading were core minor domains.
- Cambodia, Dushanbe (Tajikistan), the Kurdistan Region (Iraq), Mauritius, Rwanda and Zambia used the paper-based version of the assessment.
- The OECD advises caution when interpreting estimates for Albania, Canada, the Netherlands, New Zealand, Norway and the United States because one or more sampling standards were not met.
- Percentages shown in country notes are rounded. Small differences between rounded percentages should not be treated as evidence of a statistically detectable difference.
- Country/economy names follow OECD PISA 2025 reporting terminology, including B-S-J-Z (China), Macao (China) and Chinese Taipei.
Frequently Asked Questions
What is the minimum PISA proficiency level?
The OECD uses Level 2 as the baseline proficiency level in science, mathematics and reading. Students below Level 2 are described as low performers in the relevant subject.
What are PISA top performers?
Top performers are students reaching proficiency Level 5 or Level 6. In PISA 2025, the OECD-average top-performer shares were 7% in science, 8% in mathematics and 6% in reading.
Which subject had the largest low-performer share in the OECD average?
Mathematics. About 35% of students across OECD countries were below Level 2 in mathematics, compared with 31% in reading and 26% in science.
Which country or economy had the highest Level 2+ shares?
B-S-J-Z (China) recorded 97% at Level 2 or higher in science, 96% in mathematics and 93% in reading. It also had the largest mathematics top-performer share reported in the OECD country notes, at 54%.
Are proficiency percentages the same as PISA scores?
No. A mean score summarizes average performance on a PISA scale. A proficiency percentage reports the share of students whose performance falls above or below defined thresholds on that subject’s scale.
Sources and Measurement
Student-performance chapter covering proficiency, score distributions and PISA 2025 comparisons.
OECD: Student performance in PISA 2025Official country/economy tables. Proficiency data are reported in Table I.B1.2a.10 for science, I.B1.2a.11 for reading and I.B1.2a.12 for mathematics.
OECD: Results for countries and economiesOfficial OECD database, documentation and public-use information for the 2025 assessment.
OECD: PISA 2025 DatabaseInteractive OECD presentation of mean performance, proficiency and related PISA 2025 indicators.
OECD: Student ScoresOfficial definitions of the science, mathematics and reading domains assessed in PISA 2025.
OECD: PISA 2025 Assessment and Analytical FrameworkData release: 8 September 2026. Country-note percentages in the comparison table are rounded to whole percentages. For exact level-by-level distributions and standard errors, use OECD Tables I.B1.2a.10–12 and the associated PISA 2025 database files.