PISA 2025 results (Volume I): Future-ready students
The PISA 2025 results highlight a significant decline in global student performance, particularly in reading and mathematics. This volume examines the proficiency of 15-year-olds across 91 systems, exploring how digitalisation, socio-economic factors, and attitudes towards learning and the environment shape student readiness for an AI-integrated future.
Please login or join for free to read more.
OVERVIEW
1. Multidimensional student learning in PISA 2025
In 2025, more than 760 000 students, representing around 33 million 15-year-olds across 91 countries and economies, participated in the PISA assessment. This ninth cycle focused on science, alongside reading, maths, and an innovative assessment of learning in the digital world. The results reveal an uneven picture where, following the drop in performance recorded in PISA 2022, learning outcomes have largely stagnated or continued to decline. Across the OECD, 20% of 15-year-old students are now estimated to be low performers across all three subjects, an increase from 16% in 2022. These students lack the foundational skills needed to succeed in further education, employment, and everyday life. PISA results provide a compass for policymakers, connecting outcomes with data from students, teachers, and schools to understand why performance differs. The framework recognises that environments and relationships are intertwined with learning, and policymakers are encouraged to favour depth over breadth by teaching a smaller set of core topics well for genuine mastery.
2. Student performance in PISA 2025
OECD average scores in science fell from 489 to 482 points between 2015 and 2025. In maths, average scores fell by 22 points, equivalent to just over one year of learning. Reading recorded the steepest decline, falling by 28 points across OECD countries and 16 points among non-OECD participants. B-S-J-Z (China) and Singapore were the highest-performers across all three domains. In computational problem solving, Macao (China) achieved the highest score of 572 points, followed by Singapore at 563 points. Good policies are stronger predictors of performance than spending; Luxembourg is the biggest spender at USD 288,521 per student but posted lower science scores than Ireland, France, and Italy, which invested less than half that amount. Socio-economic background remains strongly associated with student achievement, accounting for 12% of within-country variation in science performance. On average, 37% of disadvantaged students were low performers in science, compared with 12% of advantaged students.
3. Student attitudes towards learning
This volume examines motivation, agency, and learning-related attitudes. On average, 57% of students do not find learning new things boring, though this share decreased in 60 systems since 2022, with over 10-percentage-point decreases in 30 systems. Around 60% of students reported putting in extra effort when work becomes challenging. Girls generally reported higher levels of curiosity, perseverance, and growth mindset, while boys reported higher cognitive adaptability. Across the OECD, 69% of students reported a growth mindset. Most troubling is the decline in critical skills; the share of “hasty readers” who rush through texts for quick, incorrect answers almost doubled between 2018 and 2025. Education systems with larger declines in reading performance also tended to show larger declines in student engagement. Curiosity to learn remains positively related to science performance in all participating systems with available data.
4. Student school life and beyond
Digital and AI tools are increasingly embedded in learning, but their relationship with performance depends on usage. Moderate use of digital devices for learning at school is associated with stronger science performance, while excessive use and digital distraction are associated with weaker outcomes. On average, 28% of students reported that their peers are distracted by digital devices during most or every science lesson. Schools with clear policies on mobile phone use report lower levels of distraction, and policymakers are advised to limit non-instructional device use during lessons. AI for schoolwork is widespread, but regular use was reported by only one in five students. Non-users of AI chatbots for specific tasks currently outperform those who do. Schools can help students use AI more critically, but these learning opportunities are reported more frequently by socio-economically advantaged students. Teachers must be equipped with strong skills to manage classrooms and keep students engaged, while resources should be matched to the specific needs of disadvantaged schools.
5. Students and the environment
Student environmental agency emerges from an interplay of knowledge, beliefs, and action. On average, 74% achieved at least baseline proficiency (Level 2) in environmental science. Higher-performing students tend to report more positive environmental awareness and belief in collective efficacy. Over 90% of students in Viet Nam and Indonesia are interested in contributing to environmental protection through their future careers, compared to an OECD average of 62%. Classroom opportunities for environmental learning are consistently associated with stronger pro-environmental attitudes and commitment. However, environmental awareness declined across all five topics assessed between 2015 and 2025. The largest drop was observed for nuclear waste awareness, which fell by 9 percentage points. Building strong scientific skills and attitudes is essential for sustainable decision-making and future technological breakthroughs.