TOEIC scores explained: 10–990, ETS's CEFR cut scores, and why a 40-point gain is real

200 multiple-choice questions in about two hours. Listening is 100 questions in about 45 minutes, in four parts: photographs (6), question-response (25), conversations (39) and talks (30). Reading is 100 questions in 75 minutes: incomplete sentences (30), text completion (16), single passages (29) and multiple passages (25).

Your number of correct answers is converted to a scaled score from 5 to 495 for each section, and the total is the two added together, so 10 to 990. Scores move in steps of five.

The conversion table is secret, on purpose. ETS builds a separate table for every test form by statistical equating, and because forms are reused around the world it releases neither the forms nor their tables. Our reading: any website that turns "80 correct" into a TOEIC score is guessing; ETS itself says the same raw score can give different scaled scores on different forms.

ETS asked a panel of 22 experts from ten European countries for the minimum TOEIC score at each CEFR level. Its published table, listening then reading:

A1 · listening 60 · reading 60

A2 · listening 110 · reading 115

B1 · listening 275 · reading 275

B2 · listening 400 · reading 385

C1 · listening 490 · reading 455

C2 · no score in either section

Two things in that table are easy to miss. C1 in listening is 490 of a possible 495, five points from the top of the scale. And the C1 reading score was recommended by only 45% of the panel, below the two-thirds ETS set as its own threshold; ETS publishes it with that footnote. ETS calls all of these guidelines and says it does not recommend rigid cut scores.

This comparison is ours: B2 in both sections works out at a total of at least 785, but only if each section clears its own minimum. A total of 785 made of 450 listening and 335 reading is B2 in listening and B1 in reading.

ETS gives the standard error of measurement as about 25 points per section: a listening score of 300 means your true score is between about 275 and 325, 68% of the time.

For two sittings, the figure that matters is the standard error of difference, about 35 points per section. ETS's own example: a listening score that goes from 300 to 340 is not very likely to be chance, so it suggests a real improvement. Our reading: a gain of 20 or 25 points in one section may well be noise; plan practice to move a section by more than 35.

If you think a score is wrong, you can ask the local ETS office for a rescore within three months. If they find a discrepancy, they pay; if not, you may be charged a small fee.

ETS's 2025 Report on Test Takers Worldwide lists mean scores by the test taker's native country, for the 37 countries with more than 500 test takers. The highest mean totals are Germany 851 and Lebanon 850; the lowest is Indonesia 461. Japan averages 564 and Korea 682.

In every one of the 37 countries, the reading mean is lower than the listening mean. Germany's is 453 listening and 398 reading; Japan's 313 and 251. Our reading: for most test takers the reading section, not listening, is where the next 50 points are.

The same report says 33% of test takers had already taken the TOEIC three or more times, and 28% took it for a job application. ETS notes that most respondents are from Asia, so the figures may not represent other regions.

The local ETS office (EPN member) publishes when results come out; the handbook gives 7 to 21 days depending on local policy. A score from a public session is yours and can be used worldwide; one taken through an employer is an institutional score, valid only inside that organisation.

The handbook states no expiry for the score itself, but a score report will not be re-issued once two years have passed, because, ETS says, English skills may improve or decline over time, and test centres need not keep data longer.

How the TOEIC compares with the tests used for study and migration is in our guides to TOEFL, IELTS and the Duolingo English Test. All five tests side by side: IELTS, TOEFL, PTE, Duolingo and TOEIC compared.