PISA 2025: Romania Is Further Behind Estonia Than Kenya Is Behind Greece

The gap separating Romania from the top of the EU is difficult to grasp — and uncomfortable to compare

In science, the main domain assessed in PISA 2025, Estonia scored 527 points and Romania 425, a gap of 102 points. To understand the scale of that difference, consider that Greece and Kenya are separated by 99 points. In other words, the distance between Romania and Estonia in this assessment is roughly the same as the distance between Greece and Kenya.

The comparison between Israel and the Palestinian Authority is no more reassuring: they are separated by 83 points, less than Romania and Estonia. These are uncomfortable comparisons, especially for anyone inclined to see the differences between European education systems as relatively modest gaps that can easily be closed.

The OECD uses approximately 20 points as an indicative measure of the progress students typically make during one year of schooling. It would be tempting to divide 102 by 20 and conclude that Romania is five years behind. Such a calculation, however, tells us neither how many years of education Romanian students are actually missing nor how quickly the gap could be closed. The benchmark helps us understand the scale of the distance; it does not provide a timetable for catching up.

What PISA actually measures

PISA does not measure the full breadth of a child’s knowledge, imagination, artistic talent or all the qualities that a good school should nurture. It does, however, assess abilities that are difficult to dismiss as irrelevant: whether a 15-year-old can understand a text, use mathematics in a practical situation and apply what they know to solve an unfamiliar problem.

Participants are selected according to OECD methodology. School principals and teachers do not choose the students who will represent their schools, and the assessment targets an age group rather than a particular grade. The sample may therefore include high schools, national colleges and lower-secondary schools. The results reflect the abilities students have accumulated by this age over several years of education.

“The Romania that rejects PISA”

There are enough voices in Romanian education arguing that PISA tests and their results are largely irrelevant to any assessment of the country’s school system. As if functional illiteracy or a 15-year-old’s ability to compare a price in euros with one in lei had no connection to the quality of the education they have received.

Nor is the claim that “our students are taught different things” particularly convincing. The OECD examined how closely the assessment corresponded to national curricula. In reading, Romania recorded an alignment of more than 94% between PISA items and the standard curriculum, according to the responses provided by Romanian experts.

We can discuss the limitations of the assessment and everything it leaves unmeasured. But if a student cannot use what they have learned in a situation that does not resemble a familiar classroom exercise, the problem also concerns the school. Rejecting the test does not make that problem disappear.

How much does a child’s family determine the result?

Romania is among the European countries where school performance is most strongly associated with a family’s socio-economic circumstances. The school system therefore has a limited ability to compensate for the inequalities children bring with them from home.

Seen from this perspective, one of Romania’s favourable indicators becomes rather less reassuring: Romanian students perceive their parents as highly interested in their education. Some of that involvement may arise precisely from the belief that a child’s future depends too heavily on what the family can do for them.

Parental interest is valuable. It becomes a problem for the education system when a child’s prospects depend so heavily on the time, education and financial resources available to the adults at home.

“The system is poor, but my child’s school is elite”

Many parents accept that Romanian education performs poorly overall while remaining convinced that their child’s school is one of the exceptions. A comparison with nearby schools may support that belief. An international comparison is less reassuring.

Using the available microdata, we calculated the combined results of the ten best-performing schools among those sampled in Romania’s largest cities. In computational problem solving, only six points separate this group from the average student in Estonia. In reading, the distance between Romania’s urban peak and the averages achieved by Europe’s strongest education systems is similarly small.

What Romania regards as the top of its education system lies very close to what some European countries achieve as a national average. For a parent who carefully selected a school, invested time and money and sees their child achieving good results, this may be the most uncomfortable finding in the entire ranking.

What, then, remains of the claim that “Romania leads in computer science”?

At the national level, Romania scored 442 points in computational problem solving, compared with Estonia’s 541. This result tells us how widely these abilities are distributed among students. The achievements of a small number of Olympiad winners cannot answer on behalf of an entire generation.

We are not talking about children who have never seen a computer or used technology. Familiarity with devices can nevertheless conceal an inability to use them to understand something new, test a possible solution and correct it when it fails. Digital exposure does not automatically develop the ability to solve problems.

This distinction also matters for the debate over AI in schools. A student can obtain a well-written answer without knowing whether it is correct and without being able to explain how it was reached. If the introduction of AI into education is judged mainly by access to tools and frequency of use, we risk confusing the use of technology with learning once again.

Poor results, strong confidence in classroom performance

Romanian teachers evaluate their own work far more favourably than student performance might suggest. In TALIS 2024, 71% reported that they achieved their lesson objectives “quite a bit” or “a lot” across all seven areas examined, compared with an OECD average of 44%.

School principals, in turn, have a very favourable view of teachers’ digital preparedness. An OECD report on education in Romania found that almost all principals considered their teachers to have the skills required to integrate digital devices into teaching.

These responses do not mean that teachers are unconcerned about their students or that they do not work hard. They do, however, raise a question about the standards by which success in the classroom is judged. If the objectives are largely being met and teachers are considered well prepared, how do so many children reach the age of 15 without being able to use what they have learned?

Parents may be satisfied with the school they chose, teachers with the lessons they delivered and principals with the preparedness of their staff. The PISA results show how little we can rely on these assessments alone. The quality of a sc