What the New International PISA Scores Actually Tell District Leaders

American 15-year-olds just posted their worst international reading scores in a quarter century, though it was also one of their best relative rankings in decades.

Untangling how this happened is the real lesson of the most recent international test results for district leaders who have to explain them to school boards, parents, and teachers.

The Programme for International Student Assessment, or PISA, is administered by the Organization for Economic Co-operation and Development. It tests what 15-year-olds can do in reading, math, and science and includes 760,000 students in 91 countries. Its 2025 results report that U.S. reading scores fell 14 points since 2022, near a 25-year low. Math and science held roughly flat.

Because most other countries fell by similar or larger margins, the United States actually rose in the international rankings. That makes it a good-news story built on a bad-news backdrop.

The most troubling finding for district leaders is that the gap between America's highest- and lowest-performing students is now among the widest of any country tested. Where that gap narrowed at all, it was because higher-performing students lost ground, not because struggling students gained it.

PISA is not a curriculum test. It asks 15-year-olds to apply what they know to real-world problems, including reading for meaning across multiple texts, using math to reason about unfamiliar situations, and evaluating scientific claims. Scores are set so the OECD average lands near 500, and roughly 30 points is considered equivalent to a year of schooling. That puts the U.S. reading decline at something like six or seven months of lost ground.

The ranking movement is the part likely to dominate headlines, so it's worth being precise about it before anyone in a district office repeats it uncritically. The U.S. didn’t improve. It held closer to flat than most of its peers, several of whom, including longtime high performers, posted sharper declines.

Reporting from Chalkbeat captured this precisely. “The U.S. has steadily moved up largely because other countries’ scores have deteriorated faster than ours over the last decade.  A bevy of data points show American students are learning less than they used to.” Seventy-four percent of American 15-year-olds reached basic reading proficiency, against a 69 percent OECD average. That’s a number that looks reassuring until you remember it's also near the lowest reading proficiency the U.S. has recorded in this assessment's history.

The number that should actually occupy district leaders is the equity gap. The spread between the U.S.'s strongest and weakest students is now one of the largest recorded by PISA in any participating country. It’s more than 290 points in science alone, a difference on the order of several years of schooling separating students sitting in the same country's classrooms.

And the OECD's own analysis adds an uncomfortable footnote. Where socio-economic achievement gaps narrowed anywhere in the world between 2022 and 2025, it was mostly because advantaged students' scores fell, not because disadvantaged students' scores rose. That’s not a story of catching up. It’s a story of leveling down.

What's causing any of this remains genuinely unsettled. The honest answer is that nobody claims to fully know.

Education Week's reporting noted several potential reasons, including a near-doubling in “hasty reading” since 2018. That’s where students move through texts quickly but answer inaccurately, alongside declining stamina for longer, more demanding passages. Another potential factor is the pandemic-era disruption to elementary and middle school years, which remains a live suspect for today's 15-year-olds. There’s also the steady rise in screen time and the displacement of reading for pleasure.

Moreover, artificial intelligence gets a share of the blame in early commentary. But the OECD's own findings are more careful on this topic than the headlines suggest. Students who said they didn't use AI tools to draft writing assignments outperformed those who did. But among frequent AI users, those who'd also been taught how to evaluate AI-generated information performed better than those who hadn't.

As OECD Secretary-General Mathias Cormann put it, “Technology and AI can strengthen learning and help prepare young people for the future, but only when used purposefully and not as a substitute for attention, effort and understanding.” The takeaway for schools isn't that AI is the culprit. It’s that unguided AI use and guided AI use are producing different students.

None of this yields a tidy prescription for what district leaders should do. Pretending otherwise would misrepresent what the data actually show. What district leaders can usefully do is watch three things as the year unfolds.

First, examine how their state assessment data compares to this national gap-widening pattern.

Second, assess how AI tools are introduced into ELA classrooms and whether students are getting any instruction in how to use them critically.

Third, determine if local reading interventions reach the students furthest behind or mainly serve those already closer to proficient.

The PISA results don't answer those questions. They just make clear how much rides on asking them.

Bruno V. Manno is a senior advisor at the Progressive Policy Institute and a former U.S. Assistant Secretary of Education for Policy. He can be reached on LinkedIn.