A TOEIC 990 is the maximum score: a perfect 495 in Listening plus 495 in Reading, at the very top of CEFR C1. Because the score is scaled through statistical equating rather than a raw percentage, on many forms you can miss a small number of questions and still round to 495 in a section, but the cushion is not published and changes by form, so the only reliable plan is to aim for zero errors.
It is hundreds of points above the national means of 564 (Japan) and 682 (South Korea) in ETS's 2025 report. The hard part is not learning more English; it is removing your last recurring mistakes and reading fast enough to finish Part 7 with time to spare.
What's in this guide
What a 990 actually is
990 is the ceiling of the TOEIC Listening & Reading scale: 495 in Listening and 495 in Reading, added together. Neither section can go above 495, so 990 is the arithmetic maximum, not an average or a percentile. On ETS's CEFR mapping the very top of the scale sits in C1, the "fluent, nuanced, near-native" band that begins at 945.
In Japan the perfect score has its own name, 満点 (manten), and it carries real prestige, the kind of line that stands out on a résumé and in study communities. It is worth being clear-eyed, though: for almost every practical purpose, the score that changes outcomes is 785 (B2) or 860, not 990. A perfect score is mostly a personal summit.
| Score | CEFR | What it signals |
|---|---|---|
| 785 | B2 | The common hiring and admissions bar |
| 860 | B2 (top) / A | Standout; international and leadership tracks |
| 945 | C1 | Advanced, near-native range begins |
| 990 | C1 (max) | Perfect score, the ceiling of the test |
Can you get questions wrong and still score 990?
Often, yes, but you cannot count on it. The TOEIC total is not "questions right out of 200 times five." Each section is converted to a scaled score from 5 to 495 through statistical equating, a process that adjusts for the small difficulty differences between test forms so the same ability earns the same scaled score regardless of which form you sat.
A side effect of equating is a thin cushion at the top: on many forms, missing one or a couple of questions in a section still rounds up to 495, and the margin tends to be slightly more forgiving on Listening than on Reading. But ETS does not publish the raw-to-scaled table, and the cushion changes from form to form, so there is no rule like "you can miss 3 and still get 990." The only safe target is a clean, error-free run.
Do not study to "miss two or fewer." Study to miss zero, and treat any cushion as a bonus on the day. A plan built around an unpublished margin is a plan built on luck.
How rare is a perfect score?
990 sits hundreds of points above the typical test taker. ETS's 2025 Report on Test Takers Worldwide puts the mean total at 564 in Japan and 682 in South Korea. A 990 is 426 points above Japan's average and 308 above Korea's.
Here is 990 against those two national means and the C1 line:
What percentage of people actually score a perfect 990? Here honesty matters: ETS does not publish it. The official percentile tables are by section, not for the total, and they top out well before a perfect run. Anyone quoting you an exact "top 0.X%" for a 990 total is estimating. What is certain is that it is a small, elite fraction.
Why the last 100 points are the hardest
Below 800, points come from broad improvement. Above 900, points come from removing specific, repeating errors. A 900-level test taker already answers the overwhelming majority correctly, so the road to 990 is not "get better at English", it is "stop making these four mistakes, and read faster."
The national section data explains where the resistance is. Japan's averages are 313 Listening versus just 251 Reading, and Korea's are 377 versus 305. Reading trails Listening almost everywhere, and for perfect-score seekers the Reading 495, above all the speed and precision of Part 7, is usually the last wall.
What actually separates 950 from 990
- Speed, not comprehension, in Part 7: finishing with a buffer to re-check, not racing the clock
- Fine grammar and collocation in Part 5: the one distractor that is grammatically plausible but idiomatically wrong
- Indirect answers in Part 2: responses that answer a question without the expected words
- Sustained attention in Part 3 and 4: not losing a single detail across long talks
See exactly where your last points are hiding
Take a free, full-length mock and get your Listening and Reading scores separately, plus an error breakdown, so you know whether your 990 gap is speed, grammar, or attention.
Get my section breakdown →The plan for 990
A 990 campaign is an error-log project. You are not adding knowledge; you are hunting down your own recurring mistakes until they stop happening under real timing. The shape that works:
1. Sit full, strictly timed mocks, only
At this level, untimed practice lies to you. Almost anyone aiming for 990 can get every Reading question right with unlimited time; the test is whether you can do it in 75 minutes with margin. Every practice Reading section should be timed, and ideally finished with five minutes to spare.
2. Log every miss by exact cause
After each mock, write down each mistake and its single root cause: misread the question, unknown collocation, ran out of time, lost focus, careless. A 990 is really a war against your top two causes. Most 950-plateau test takers are losing points to the same two or three patterns every time.
3. Build speed in Parts 5 and 6 to fund Part 7
Every second saved in the short grammar and text-completion parts is a second banked for the long Part 7 passages, where the perfect-score points actually live. Drill Parts 5 and 6 until they are near-instant, then spend the saved time re-checking Part 7 double-passage and triple-passage sets.
4. Protect Listening; it is your likely 495
For most people Listening is the more forgiving 495. Keep it sharp with Part 3 and 4 prediction drills, reading the questions before each clip, so a single missed detail never costs the perfect section.
The traps that cost 990-seekers points
The same handful of items decide most near-perfect runs. Knowing them in advance is half the defence.
| Where | The trap | The fix |
|---|---|---|
| Part 2 | Indirect response ("Where's the report?" answered with "Mia took it home") | Expect answers that solve the problem, not echo the words |
| Part 5 | Two grammatically valid options; only one collocates | Read for natural word pairs, not just grammar |
| Part 7 | Inference and "NOT true" questions in triple passages | Cross-reference all sources before answering; watch the clock |
| Part 3/4 | A detail changes late in a long talk | Keep tracking to the end; do not lock an answer early |
None of this needs a cram school. It needs full-length practice in the real format, ruthless error logging, and speed. That is exactly what our free mocks and AI coach are built to give you, the coach sorts your misses by cause automatically and rebuilds your drill list around the two that cost you most.
Start your 990 push today, free
Full-length TOEIC mocks in the official 2026 format, instant Listening + Reading scores, and an AI coach that turns your error log into a focused plan.
Take a free mock test →Frequently asked questions
What is a TOEIC 990?
Do you need every question correct to score 990?
How rare is a TOEIC 990?
Why are the last 100 points so hard?
How long does it take to go from 900 to 990?
Is a TOEIC 990 worth chasing?
Sources
- ETS, 2025 Report on Test Takers Worldwide, TOEIC Listening & Reading. Mean totals: Japan 564 (L 313 / R 251), South Korea 682 (L 377 / R 305).
- ETS, TOEIC 4-Skills Tests CEFR Mapping Table. C1 band begins at 945; 990 is the scale maximum.
- ETS, TOEIC Listening & Reading Percentile Rank. Section-level percentiles only; no published total-score percentile for 990.
- ETS, How the TOEIC is scored. Scores are produced by statistical equating, not a raw-to-scaled percentage; the conversion table is not published.