PISA 2025 results explained: what the OECD findings mean for teachers

What PISA is and what the 2025 cycle measured

The OECD’s Programme for International Student Assessment (PISA) tests what 15-year-olds know and can do, and how well they apply it to unfamiliar problems. Questionnaires for students and principals add context about attitudes, school life and learning habits. Results are reported for education systems, not for individual students or schools.

PISA 2025 made science the major domain, with mathematics and reading as minor domains. It also introduced Learning in the Digital World, an assessment of computational problem solving and self-directed learning with digital tools. More than 760 000 students took part, representing about 33 million 15-year-olds in 91 countries and economies. Volume I was published on 8 September 2026. The OECD has said results from the optional Foreign Language Assessment will follow in April 2027, with fuller Learning in the Digital World results in May 2027.

The headline numbers

According to the OECD, PISA 2025 recorded the lowest OECD-average performance observed so far in science, reading and mathematics. Reading has declined since peaking around 2012. Science has fallen more gradually over the past decade and was stable between 2022 and 2025. Mathematics has fallen especially sharply since 2018.

Between 2022 and 2025 alone, the OECD average fell by 14 points in reading, 9 points in mathematics and 3 points in science. The share of students who are low performers in all three subjects reached 19.7%.

B-S-J-Z (China) and Singapore were the highest performers across science, mathematics and reading. Estonia, Japan, Korea, Macao (China), Chinese Taipei and the United Kingdom were also among the ten highest-performing systems in all three subjects. In computational problem solving, Macao (China) scored highest at 572 points.

Reading: the rise of the “hasty reader”

The OECD highlights a decline in the very reading skills that matter most when information is abundant: evaluating information, connecting ideas across several sources and thinking critically about a text. The share of “hasty readers”, students who rush through texts and give quick but incorrect answers, almost doubled between 2018 and 2025. We explore classroom responses in Hasty readers: teaching slower, deeper reading.

Which systems improved?

Not every trend points downwards. Among OECD countries, science scores rose after 2022 in Costa Rica, the Slovak Republic, Türkiye and the United Kingdom. Elsewhere, improvements were seen in Cambodia, El Salvador, Georgia, Jordan, Mongolia, Montenegro, the Philippines, Saudi Arabia, Thailand, the United Arab Emirates, Uruguay and Uzbekistan.

Money is not the whole story. The OECD notes that national wealth accounts for only about half of the variation in countries’ performance. Luxembourg spends by far the most per student, yet posted lower average science scores than Ireland, France and Italy, each of which spent less than half as much.

Equity: the gap narrowed for the wrong reason

Across the OECD, socio-economically advantaged students scored 85 points higher in science than disadvantaged students, and socio-economic status explained around 12% of the variation in science performance within countries. The gap narrowed between 2022 and 2025, but mostly because advantaged students performed less well, not because disadvantaged students improved.

Some systems combine strong performance with fairness, including B-S-J-Z (China), Canada, Estonia, Ireland, Japan, Korea and Macao (China). On average, around 12% of disadvantaged students were “academically resilient”, scoring in the top quarter of science performance in their own country.

Engagement and learning habits

Volume I looks closely at how students approach learning. On average, 57% of students said they do not find learning new things boring, but that share fell in 60 education systems after 2022. Around 60% said they put in extra effort when work becomes challenging. Curiosity, perseverance, help-seeking, goal setting and a growth mindset are positively associated with science performance, and systems with larger falls in reading also tended to see larger falls in engagement. See Metacognition and self-regulated learning after PISA 2025.

Digital devices and AI

Limited or moderate use of digital devices for learning at school is often associated with better outcomes. When devices are used for leisure at school, that relationship reverses sharply. On average, 28% of students reported that classmates are distracted by devices in most or every science lesson. Schools with clear phone policies, whether bans or well-defined guidelines, report less distraction. Read more in Phones, devices and distraction.

The OECD describes the relationship between AI use and performance as complex. Students who do not use AI for specific schoolwork tasks outperform those who do, while weekly users and non-users of AI for general purposes perform similarly. Students tend to perform slightly better when frequent AI use is paired with opportunities to develop AI literacy, but those opportunities are more common among advantaged students. We unpack this in What PISA 2025 says about students using AI chatbots.

Classroom climate and support

There are encouraging signals. Students’ sense of belonging improved in 39 systems after 2022. Disciplinary climate in science lessons improved between 2015 and 2025 in 37 systems, which is associated with stable or improved science performance. In half of the participating systems, students reported more teacher support in science lessons than in 2015. Study help such as peer-to-peer tutoring is associated with higher science performance. Students also reported less family support than in 2022 in almost every system.

What this means for teachers

PISA does not evaluate individual teaching methods, and associations in the data are not proof of cause and effect. Even so, the 2025 findings point towards several classroom priorities:

• Protect deep reading. Plan time for sustained reading of longer texts and for comparing several sources, not only for extracting single facts.
• Teach AI literacy explicitly. Show students how to question, verify and improve AI-generated content, and make sure every student gets that opportunity, not only those who learn it at home. Start with our AI literacy classroom activities.
• Make device use purposeful. Agree when devices are needed for learning and when they are put away.
• Build learning habits. Model planning, self-questioning and asking for help, and give students time to practise them.
• Find gaps early. Use frequent formative assessment and retrieval practice to spot students at risk of falling below baseline proficiency, then differentiate and remediate.
• Use peers well. Structured peer tutoring and collaborative explanation give more students access to help.