PISA 2025 Results Explained for Parents: What Changed Since 2022

Here are the PISA 2025 results explained the way I wish someone had explained them to me: the OECD tested more than 760,000 fifteen-year-olds across 91 education systems, and on September 8, 2026 it reported the lowest average math and reading scores it has ever recorded. Reading fell 28 points across OECD countries between 2015 and 2025. Math fell 22. Science barely moved. One in five students is now a low performer in all three subjects at once, up from 16% in 2022.

Most of the coverage you have seen is a horse race: who is number one, who slipped a rung. That framing is almost useless to a parent. The number that matters is not where a country sits in a table. It is which skills are eroding, for which kids, and why.

Key Takeaways

• The OECD average fell in every core domain: science 482, reading 461, math 463. All three are the lowest values PISA has ever recorded for the OECD.
• Reading dropped 28 points in a decade. The OECD’s rule of thumb is that 15-year-olds gain about 20 points per school year, so that is roughly a year and a half of learning.
• The share of students who are low performers in science, reading and math rose from 16% (2022) to 20% (2025).
• The decline is not uniform: it shows up most on long texts and on tasks that demand evaluation and integration, and the share of “hasty readers” almost doubled since 2018.
• Several systems moved the other way (the United Kingdom, Türkiye, Costa Rica, the Slovak Republic in science), so the trend is not inevitable.

PISA 2025 results explained: what was measured and what the headline numbers say

PISA is the OECD’s Programme for International Student Assessment, a two-hour test given every three years to a representative sample of 15-year-olds in each participating system (in the United States, it is run by the National Center for Education Statistics). It measures how well students can apply science, reading, and math to unfamiliar problems, not whether they memorized a curriculum. The 2025 cycle, released on September 8, 2026, focused on science and added a new domain, computational problem solving.

The OECD’s own preface, and its September 8 press release, are blunt. “Following the unprecedented drop in performance recorded in PISA 2022, learning outcomes have largely stagnated or continued to decline in many countries.” It also notes that the weakening “predates the COVID-19 pandemic in many countries.”

Three details matter more than any rank:

The decline is long-term, not a pandemic blip. Reading peaked around 2012 and has fallen since. Science has declined gradually over the decade. Math held near its 2003 level through 2018, then dropped sharply.

It hits the bottom harder than the top. Across OECD countries, the share of students below Level 2 in reading rose 10 percentage points between 2015 and 2025, while the share at Level 5 or 6 fell only two. Math shows the same shape: about nine points more low performers, two fewer top performers.

Socio-economic gaps narrowed for the wrong reason. The OECD writes that narrowing “reflected declines among advantaged students rather than improvements among disadvantaged students.”

The 2015 to 2025 score table, translated into school years

The OECD tracks trends on a consistent set of countries, so the trend figures differ slightly from the headline averages in older reports. The table uses the OECD’s trend basis from Volume I. To convert points into time, the report’s Box I.2.2 estimates that students around age 15 typically gain about 20 score points over one year of schooling.

Domain	OECD avg 2015	OECD avg 2025	10-year change	2022 to 2025 change	Roughly in school years
Reading	489	461	−28	−14	About 1.4 years lost
Mathematics	485	463	−22	−9	About 1.1 years lost
Science	489	482	−7	−3	About a third of a year
Computational problem solving	new in 2025	500	n/a	n/a	n/a

Two cautions. First, the “years of learning” conversion is a rough average; the OECD says country-specific estimates vary widely. Second, PISA is a cross-section of 15-year-olds each cycle, not a cohort study, so “lost” means “this year’s 15-year-olds score lower than 2015’s,” not that any individual child regressed.

Why reading fell the most, according to the OECD’s own analysis

Volume I includes a long box on what could explain the reading decline. It refuses to pick a single villain, and that restraint is worth copying. The OECD frames four levels of explanation as “complementary” rather than rival.

At the student level, the drop is not uniform across tasks. Success rates fell fastest on items with long texts and on items requiring students to evaluate and reflect, “which require careful reading, remembering and integrating multiple pieces of information.” Tasks that only asked students to locate a single fact held up best. The fluency items at the start of the test tell the same story: the share of accurate, fluent readers fell about seven percentage points between 2018 and 2025, while “hasty readers” who answer fast and wrong rose from almost 7% to 11%.

At the school level, TALIS 2024 data show teachers spending 16% of class time keeping order in 2024, up from 13% in 2018. Bullying, which had fallen between 2018 and 2022, rose again by 2025.

At the societal level, more students reported spending over an hour a day on social networks or games, while fewer reported using devices to look up practical information or create content. The report is careful: “These trends do not establish a direct link between online activities and declining reading performance.” It does cite outside evidence pointing the same way, including a 2023 meta-analysis in Review of Educational Research by Altamura, Vargas, and Salmerón, which pooled 40 effect sizes from almost 470,000 participants and found that leisure reading on digital devices has only a small relationship with comprehension (r = .055), far weaker than the medium-sized link for print.

And the pandemic? The OECD makes a point I found genuinely surprising. School closures in early 2020 ranged from zero weeks in Sweden to about 20 in Costa Rica and Colombia. Yet reading fell about 21 points in Sweden, about 9 in Colombia, and did not change significantly in Costa Rica. Closure length alone does not predict the damage.

If you want the longer version of the phone-versus-curriculum debate, our companion piece on the PISA 2025 reading decline and what parents can do walks through each hypothesis and how strong the evidence is.

What the engagement data add to the picture

The PISA 2025 questionnaire asked students about curiosity, perseverance, and how they feel about school. The pattern is not encouraging. On average, 57% of students say they do not find learning new things boring, but that share fell in 60 systems since 2022, by more than ten percentage points in 30 of them. No system moved the other way. The share who say they put in extra effort when work gets hard (60%) fell in 32 systems.

These attitudes track the score trends. The OECD reports that changes in the share of students who agree they are “curious about many different things” correlate with changes in reading performance across systems (r = 0.54). Correlation is not cause, and the report says so. But it does suggest that the erosion is not only about skills. It is also about willingness to sustain effort on a hard task, which is exactly what a long reading passage or a multi-step math problem demands.

Which systems bucked the trend, and what they had in common

Within the OECD, science scores rose since 2022 in Costa Rica, the Slovak Republic, Türkiye, and the United Kingdom. Outside it, gains showed up in Cambodia, El Salvador, Georgia, Jordan, Mongolia, Montenegro, the Philippines, Saudi Arabia, Thailand, the United Arab Emirates, Uruguay, and Uzbekistan.

The OECD is careful not to claim a recipe. But it names some levers in the preface: consistent expectations in the three core subjects; “depth over breadth” in curriculum; classrooms “kept free of avoidable digital distractions”; and AI used “in a targeted way for feedback and personalised practice” rather than as a substitute for effort. Its line on that last point is one worth putting on the fridge: “real understanding comes from doing the thinking, not from having a tool do it instead.”

On the AI question specifically, 46% of OECD students say they use chatbots weekly or more to help them learn. Students who use AI for specific tasks like summarizing or drafting score lower in science than non-users, while weekly users “to help me learn” perform about the same as non-users once socio-economic status is accounted for. Those are associations. We cover the full U-shaped pattern in why moderate AI use beat heavy and zero use in PISA 2025.

What to actually do at home

None of this requires a tutor or a new app. The skills that eroded most are the ones that can be rebuilt with boring, repeatable habits.

Protect one long read a week

The steepest declines were on long texts. Pick one piece per week that is longer than your kid would choose on their own: a magazine feature, a chapter, a long-form article. Read it in one sitting, then ask one question that requires connecting two parts of the text. That is the exact skill PISA says is fading.

Make “slow and right” beat “fast and wrong”

Hasty readers nearly doubled. When your child rushes homework, do not reward speed. Ask them to explain one answer out loud before they move on. If they cannot, the answer was a guess. Ten minutes of this a few times a week changes the default.

Keep phones out of the homework room

The OECD found that in schools with clear phone policies, fewer students report distraction, and 28% of students say classmates are distracted by devices in most or every science lesson. A phone in another room during homework is the cheapest intervention on this list.

Ask about effort, not grades

Since the share of students who “finish what they start” fell in 40 systems, make persistence the thing you notice. “What was the hardest part today and what did you do when you got stuck?” gets more useful information than “What did you get?”

What not to do

Do not turn PISA into a reason for a homework crackdown. The report’s own analysis says engagement and performance fell together; adding pressure without adding interest tends to buy compliance, not curiosity. And do not treat the ranking as a verdict on your child’s school. A country average says nothing about the classroom your kid sits in.

What to Watch For Over the Next 3 Months

• Week 4: Your child can read a 1,500-word piece in one sitting without checking a device, and can summarize it in three sentences.
• Month 2 red flags: Homework answers that are consistently fast and wrong, or a kid who says “it’s boring” about everything new. Both showed up in the PISA data as leading indicators.
• Month 3 self-check: Ask whether your child chose something long to read on their own at least twice this month. If not, adjust the supply (library trip, a series they like), not the pressure.

Frequently Asked Questions

What does a PISA score of 461 in reading actually mean?

PISA scales are set so the OECD average was 500 when each domain was first assessed, with a standard deviation of about 100. A score of 461 means the average 15-year-old in OECD countries now reads roughly 40 points, or about two school years by the OECD’s rule of thumb, below that original benchmark. It is a relative measure, not a percentage correct.

Did PISA 2025 change the test, and is that why scores fell?

No. The core reading, math, and science scales are linked across cycles so that trends are comparable. The one new element, computational problem solving, has its own scale and does not affect the other three. The OECD does flag sampling caveats for some countries, including the United States, which is why those appear with an asterisk.

Is the decline mostly a COVID effect?

Partly, but the OECD says the weakening “predates the COVID-19 pandemic in many countries.” Reading peaked around 2012. The pandemic likely accelerated the trend and may have shifted habits, but closure length does not predict how far a country fell.

Should I worry if my child is in a top-performing school?

The decline hit the bottom of the distribution harder than the top, so a strong student in a strong school is less exposed. But engagement fell broadly, including among advantaged students. The habits above cost nothing and help everyone.

Where do US students fit in this?

US math (463) sits at the OECD average, reading (490) and science (502) are above it, and computational problem solving (513) is above the 500 average. The US also carries an asterisk because its school response rate was 54% after replacement. We break that down in why US rank rose while scores stayed flat.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. OECD. (2026). “PISA 2025 Results (Volume I): Future-Ready Students.” OECD Publishing. https://www.oecd.org/en/publications/pisa-2025-results-volume-i_73451bc5-en.html
2. OECD. (2026). “PISA 2025: Students’ reading and mathematics performance declined sharply across the OECD.” OECD press release, September 8, 2026. https://www.oecd.org/en/about/news/press-releases/2026/09/pisa-2025-students-reading-and-mathematics-performance-declined-sharply-across-the-oecd.html
3. OECD. (2025). “TALIS 2024 Results.” OECD Publishing. https://www.oecd.org/en/about/programmes/talis.html
4. National Center for Education Statistics. (2026). “Program for International Student Assessment (PISA) 2025.” NCES. https://nces.ed.gov/surveys/pisa/
5. Barnum, M., & Meltzer, E. (2026, September 8). “PISA results 2025 show U.S. gaining as other countries decline.” Chalkbeat. https://www.chalkbeat.org/2026/09/08/pisa-results-2025-show-us-gaining-as-other-countries-decline/
6. Altamura, L., Vargas, C., & Salmerón, L. (2023). “Do New Forms of Reading Pay Off? A Meta-Analysis on the Relationship Between Leisure Digital Reading Habits and Text Comprehension.” Review of Educational Research. https://doi.org/10.3102/00346543231216463
7. Hong, J. (2026, September 11). “Students are using AI in school and test scores are down, but the OECD finds that those who use it once or twice a week perform similarly to nonusers.” Fortune. https://fortune.com/2026/09/11/students-ai-school-test-scores-moderate-use-vs-daily-users/