Romania’s Baccalaureate, 2015–2026 · about 1,400 high schools a year

Romanian high schools with a 0% Bac pass rate, and how much of every July ranking is chance

In 2026, 27 Romanian high schools had not a single graduate pass the Baccalaureate, the school-leaving exam. At 19 of them five students or fewer sat it. A pass rate from three students can only be 0, 33, 67 or 100%.

The 27 schools at 0% in 2026, one square per candidate

passedfailed

The typical school: 59 candidates

For comparison, the typical school of 2026: 59 candidates, 82% passed

What would chance alone produce? Each school keeps its 2026 number of candidates and its pass rate from 2023–2025; then each candidate passes or fails, like a roll of the dice.

The 27 schools, one by one

Technological high school1 candidate
Technological high school1 candidate
Technological high school1 candidate
Technological high school1 candidate
Technological high school1 candidate
High school2 candidates
Technological high school2 candidates
Technological high school2 candidates
Technological high school2 candidates
Technological high school2 candidates
Technological high school3 candidates
Technological high school3 candidates
Technological high school3 candidates
Technological high school3 candidates
Technological high school4 candidates
Technological high school4 candidates
High school4 candidates
Technological high school4 candidates
Technological high school5 candidates
Liceul Tehnologic „Ion Ghica” OltenițaCălărași7 candidates
Colegiul Tehnic ReșițaCaraș-Severin8 candidates
Liceul Tehnologic „Marin Grigore Năstase” TărtășeștiDâmbovița9 candidates
Liceul Tehnologic Ucecom „Spiru Haret” IașiIași9 candidates
Liceul Tehnologic Special pentru Deficienți de Auz Cluj-NapocaCluj10 candidates
Liceul Tehnologic „Aurel Persu” Târgu MureșMureș12 candidates
Liceul de Agricultură și Industrie Alimentară OdobeștiVrancea13 candidates
Liceul Tehnologic „Danubius” CorabiaOlt23 candidates

Schools with five candidates or fewer appear without a name or county. England’s school tables do not publish results at all for schools with five pupils or fewer; here they appear only as squares, to show how small they are.

In July the Romanian press reported 32 schools at 0%: that number was calculated by the Bacplus platform from results before appeals and counted differently. This page uses final results, for that year’s graduates who sat the exam only.

Any school, year by year, and how far chance can move it#

Type a school’s name or its town. The chart shows the pass rate each year; the band behind it shows where luck alone would put the rate, in 95 years out of 100, for a school with that many candidates and its usual rate.

Find a school

1,561 schools named from the school register. Or open one of these:

Every school, every year: how far each rises and falls from one year to the next#

Each dot is a school in one year: how much its pass rate changed from the year before, against how many candidates it had. The band shows how far chance alone moves a school at the national rate (78%, the 2015–2026 average).

↔ candidates that year↕ change from the year before (points)

Tap a dot for details. A pair of years is formed only for the same school code in two consecutive years. The chart leaves out 655 pairs in which only one of the years has five candidates or fewer; the share below includes them.

Schools with a handful of candidates jump from −100 to +100 points. Schools with hundreds usually stay within a few points of zero. The 20 schools that “rose” or “fell” most in a year had, at the median, three to seven candidates; the ordinary school had 44 to 54.

Solid dots (14.1% of all pairs of years) changed by enough that Fisher’s exact test gives p below 5%: hard to put down to chance alone. If nothing changed at any school, at most about 5% would still look like this, and part of the rest comes from the national rate rising after 2021. So real changes exist, but most rises and falls cannot be told apart from chance.

The top is different: 100% schools are usually large#

The top list reads differently. In 2026, 108 schools had 100%, and half of them had more than 76 candidates: colleges where nearly everyone passes every year. There, a 100% says something about the school.

Nor are small schools secretly just as good. At schools with ten candidates or fewer, 36% of candidates passed over the 12 years; at schools with more than 200, 91%. At three candidates, a 0% does not show the school is weaker than another small school. It shows that three students did not pass.

By school size, 2015–2026

Candidates	Passed	Change from year to year	Year-to-year likeness
1–10	36%	16.7 points	0.55
11–30	47%	11.4 points	0.76
31–80	61%	7.5 points	0.86
81–200	82%	3.3 points	0.92
over 200	91%	1.2 points	0.94

Candidates: how many sat the exam in a year. Change: the typical (median) shift in the pass rate from one year to the next. Likeness: the correlation between one year and the next, from 0 (none) to 1 (identical).

“The most dangerous equation”: why extreme results come from small groups#

In 1733 Abraham de Moivre showed that the average of a group scatters less and less as the group grows, with the square root of its size. A school with four times as many candidates strays by luck half as far; one with a hundred times as many, a tenth as far.

σ_mean = σ / √n

How far luck moves the pass rate of a school where 78% usually pass

In 2007 the statistician Howard Wainer called it “the most dangerous equation”, because not knowing it leads to costly decisions. The Gates Foundation funded small schools for years, partly because many of America’s best schools were small; by 2001 it had given about $1.7 billion to education projects. Wainer showed, on Pennsylvania’s schools, that small schools were over-represented among the worst 50 too: nine of them were among the 50 smallest. The equation puts small schools at both ends.

In 2009 Bill Gates wrote that the foundation had made over $2 billion in high-school grants and that many of the small schools it funded did not improve students’ achievement in any significant way. He put the failure down to how the schools were run; Wainer showed that the evidence the strategy started from was partly the size effect alone.

The same effect shows on disease maps: in the United States both the counties with the highest kidney-cancer rates and those with the lowest are mostly rural counties with few people. On any map coloured by commune or county, the first question is the same: out of how many?

A related effect, regression to the mean, explains part of the Dunning-Kruger chart.

Out of how many? A check for any ranking#

It works for a school, a hospital, a town or a firm: any percentage that went up or down. Enter how many cases there were each year and how many “succeeded”.

The starting example: a school with 3 of 7 passing one year and 0 of 6 the next.

Questions#

Which Romanian high schools had a 0% Bac pass rate in 2026?

In the final results, 27 schools had no graduate of that year pass. Of these, 19 had five candidates or fewer and are not named here. The others: Liceul Tehnologic „Ion Ghica” Oltenița (7 candidates); Colegiul Tehnic Reșița (8 candidates); Liceul Tehnologic „Marin Grigore Năstase” Tărtășești (9 candidates); Liceul Tehnologic Ucecom „Spiru Haret” Iași (9 candidates); Liceul Tehnologic Special pentru Deficienți de Auz Cluj-Napoca (10 candidates); Liceul Tehnologic „Aurel Persu” Târgu Mureș (12 candidates); Liceul de Agricultură și Industrie Alimentară Odobești (13 candidates); Liceul Tehnologic „Danubius” Corabia (23 candidates).

Why do small schools end up at the bottom of rankings and among the “biggest rises”?

Because a percentage from a few students moves a lot from year to year on luck alone. At schools with ten candidates or fewer the rate typically changes by 16.7 points from one year to the next; at schools with more than 200, by 1.2.

Does that mean a 0% school is as good as the others?

No. Very small schools do much worse on average: 36% passed, against 91% at large schools. But a 0% from three candidates cannot say whether the school is weaker than another small school at 33%.

Why 27 schools and not the 32 reported in July?

The 32 came from the Bacplus platform, from results before appeals, counted differently (the Corabia school, for example, appears there with 33 candidates, here with 23 graduates of the year who sat the exam). These are final results, after appeals; a few candidates passed after re-marking.

How is the pass rate calculated here?

Passes divided by that year’s graduates who sat the exam. Candidates from earlier years, absentees and those excluded are left out.

What is the Bacalaureat?

Romania’s school-leaving exam, taken at the end of high school, a little like A-levels in England. To pass, a candidate needs at least 5 out of 10 in each written exam and an average of at least 6.

Sources and method#

Results: the Ministry of Education’s files with every candidate’s results, June–July session, published on data.gov.ro (OGL-ROU-1.0 and CC BY 4.0 licences), 2015–2026. Only that year’s graduates (current cohort) who sat the exam are counted: passed or failed. The files hold final results, after appeals: for 2026 they count 113,249 candidates present; the ministry’s press file counts 113,296, including the 46 excluded. School names come from the school-network register (data.gov.ro, CC BY 4.0), by SIIIR code; 23 codes appear in no published register. A school that changed its code appears as two lines. For years in which a school had five candidates or fewer, neither the number of candidates nor the result is published. Test for a change from one year to the next: Fisher’s exact test, 5% threshold. Chance band: binomial 2.5% and 97.5% quantiles. The draw: each school at its 2023–2025 rate, 5,000 draws; schools without results in those years are left out.