How the Duolingo English Test is scored
The 10 to 160, in increments of 5 scale, all eight subscores, and what the adaptive engine is actually doing while you take the test.
- All eight subscores and the arithmetic that links them
- A real adaptive engine — difficulty moves with your answers
- Official CEFR, IELTS and TOEFL alignment
New to the exam? Start with what the Duolingo English Test is — the format, cost and acceptance in one place; this page is the deep dive on scoring. Almost every explanation of DET scoring stops at "it is adaptive and it goes up to 160". That is true and it is not enough to act on. This page covers the whole mechanism: what the number on the scale means, where the eight subscores come from, how the arithmetic that produces your overall score works, and what the adaptive engine is doing question by question.
Everything here about the official test comes from Duolingo's published material. Where something is our own observation from building a scoring engine, or an estimate rather than a published fact, it says so in the sentence.
Any score, and what it corresponds to
Drag the slider. Rough IELTS and TOEFL equivalents are shown alongside.
Want just the conversion on its own page? The DET score converter turns any Duolingo score into its IELTS, TOEFL and CEFR equivalent. And to understand what the number is actually built from, see Duolingo English Test scores explained.
What the number actually is
The Duolingo English Test reports on a scale from 10 to 160, in increments of five. There are no half points and no scores in between — a 117 does not exist.
It is not a percentage and it is not a raw count of correct answers. Two candidates who answered the same number of questions correctly can receive very different scores, because the test is adaptive and the difficulty of what you were asked is part of the calculation. Getting a hard question right is stronger evidence about your ability than getting an easy one right, and the scoring reflects that.
The practical consequence is one people find counter-intuitive: feeling like the test got hard is usually a good sign. The engine raises difficulty when you are succeeding. A test that felt comfortable throughout often means it settled at a lower level early.
Where the eight numbers come from
Your report carries eight subscores, and they are not eight independent measurements. Four are measured; four are derived from those four.
The four individual subscores — Reading, Writing, Speaking and Listening — are the ones the test actually estimates from your responses. The four integrated subscores — Literacy, Comprehension, Conversation and Production — are each the average of two individual ones, pairing the skills that get used together in real situations.
| Integrated subscore | Built from | What it describes |
|---|---|---|
| Literacy | Reading + Writing | Working with written English in both directions. |
| Comprehension | Reading + Listening | Taking English in, written or spoken. |
| Conversation | Listening + Speaking | Real-time spoken exchange. |
| Production | Writing + Speaking | Generating English rather than receiving it. |
This is why the integrated numbers can never surprise you if you have the individual ones. If your Reading is 115 and your Writing is 105, your Literacy is 110 — there is no separate Literacy measurement that could come out differently.
How the overall score is built
Follow the numbers up from the four measured skills.
Your overall score is the average of the four individual subscores, rounded to the nearest five — not an average of all eight, and not a separate global measurement.
Averaging all eight would double-count: the integrated scores are already made of the individual ones, so including both weights each skill twice. In the worked example above, the four individual scores are 115, 105, 105 and 125. Their mean is 112.5, which rounds to 115.
One consequence worth planning around: because the overall is a mean of four, a single weak skill drags it in a predictable way. Twenty points of weakness in one skill costs five points overall. If an institution also sets minimum subscores — many do — then that weak skill can block you even when your overall clears the bar comfortably.
What the engine is doing while you answer
The DET is computer-adaptive. Rather than serving everyone the same questions, it chooses each next item based on everything you have done so far, aiming at the questions that will tell it the most about you.
Concretely: the engine holds an estimate of your ability and a measure of how uncertain it is. It picks the item whose difficulty sits closest to that current estimate, because an item you have roughly a coin-flip chance on is the most informative one available. Your answer updates both the estimate and the uncertainty. As the uncertainty narrows, the score stabilises.
The graph above is a real convergence trace from our own engine. Notice the shape: wide swings early, when it knows almost nothing about you, then progressively smaller corrections as the green band — the confidence interval — closes in.
Three things follow from this that are worth knowing before you sit the test.
- Early questions matter more than late ones. Not because they are weighted more heavily, but because they steer the difficulty of everything that follows. A slow start puts you on an easier path that takes time to climb back out of.
- You cannot game it by pacing yourself. Deliberately answering easier or slower does not protect a score; it just gives the engine evidence of a lower ability.
- Difficulty rising is feedback. If the questions get noticeably harder as you go, the engine is telling you it has revised its estimate upward.
What your score means in a framework people recognise
Duolingo publishes an alignment between the DET scale and the Common European Framework of Reference, which is the vocabulary most universities and employers already think in.
| CEFR level | DET score | What it broadly means |
|---|---|---|
| A1–A2 | 10–55 | Basic phrases and predictable everyday exchanges. |
| B1 | 60–95 | Independent use on familiar topics; can handle most travel situations. |
| B2 | 100–125 | Can follow complex text and argue a position fluently. The band most undergraduate requirements sit in. |
| C1 | 130–150 | Effective operational proficiency. Flexible, precise, largely spontaneous. |
| C2 | 155–160 | Near-native command in essentially any register. |
The B1/B2 boundary at 100 is the one that matters most in practice, because it is the line a large share of published undergraduate requirements sit on or just above. Moving from 95 to 100 is five points on the scale and a whole band in how your result reads.
Why a single number is slightly misleading
Every test score is an estimate with error around it, and the DET is no exception. If you sat the real test twice in one week without learning anything in between, you would not get the same number twice.
That is not a flaw in the test — it is a property of measuring something as variable as language ability from one hour of evidence. What matters is knowing roughly how wide the band is, because it changes how you should read your own results.
The practical rule: treat a single practice score as weak evidence and a consistent run of them as strong evidence. Two results in a row above a threshold mean considerably more than one.
Which questions feed which subscore
If a subscore is lagging, this is the table that tells you where to work. Every task links to a full guide covering how it is scored and what loses marks.
| Question type | Time | How often | Feeds |
|---|---|---|---|
| Read & Complete | 3 minutes | 3–6 times | Reading, Writing |
| Read & Select | 5 seconds | 15–18 times | Reading |
| Fill in the Blanks | 20 seconds | 6–9 times | Reading, Writing |
| Listen & Type | 1 minute | 6–9 times | Listening, Writing |
| Speak About the Photo | 90 seconds | at least once | Speaking |
| Interactive Speaking | 35 seconds each | 6–7 questions | Speaking, Listening |
| Write About the Photo | 1 minute | at least twice | Writing |
| Interactive Reading | 7 minutes | 2 sets | Reading |
| Interactive Listening | 6 minutes | 2 sets | Listening, Speaking |
| Speaking Sample | 3 minutes | once | Speaking |
| Writing Sample | 5 minutes | once | Writing |
Two tasks feed two individual subscores at once — Interactive Speaking and Interactive Listening — which makes them the highest-leverage places to improve if you are short of time.
Questions about DET scoring
What is a good Duolingo English Test score?
It depends entirely on where you are applying, which is the honest answer even though it is not the satisfying one. As a rough map: 100–125 is B2 and covers the most common undergraduate requirements; 130–150 is C1 and covers selective postgraduate programmes. The only number that matters is the one your institution publishes, and many also set minimum subscores as well as a minimum overall. To see any score against IELTS, TOEFL and CEFR, use the DET score converter.
How is the overall DET score calculated from the subscores?
The overall score is the average of the four individual subscores — Reading, Writing, Speaking and Listening — rounded to the nearest five. It is not an average of all eight. The four integrated subscores are themselves derived from the individual ones, so including them would count each skill twice.
Why did the test get harder as I went on?
Because you were doing well. The test is adaptive: it raises difficulty when you answer correctly, since a harder question you have roughly even odds on tells it more about your level than an easy one you will certainly get right. A test that felt easy throughout usually settled at a lower estimate early.
Can I improve my score by answering slowly and carefully?
No. Deliberately pacing yourself does not protect a score — the engine estimates ability from what you produce, and slower does not read as better. On the tasks with tight timers, hesitation costs you unanswered items, which are scored as zero.
Why is my overall score lower than most of my subscores?
Because the overall is a mean, so one weak skill pulls it down more than people expect. Twenty points of weakness in a single skill costs five points overall. This is also why some institutions set minimum subscores separately — a strong overall can hide a skill that is well below their bar.
How accurate is a single DET score?
Every test score is an estimate with error around it. If you sat the test twice in a week without learning anything, you would not get an identical number. That is why our reports show a range rather than a bare point, and why a consistent run of practice scores is much stronger evidence than any single one.
Do all eight subscores appear on the official score report?
Yes — four individual (Reading, Writing, Speaking, Listening) and four integrated (Literacy, Comprehension, Conversation, Production), alongside the overall score. Our practice reports return the same eight so the shape is familiar. See a complete sample report, or read what every number on your report means.
How long are Duolingo English Test scores valid?
2 years from the test date. Institutions may impose their own, shorter recency requirements, so check the specific programme rather than assuming the full period.
Is the Duolingo English Test graded by a computer or a human?
Automatically. Reading and listening items are compared against a key; writing and speaking are scored by machine grading against a rubric. Humans are involved in proctoring review and in the research behind the scoring model, not in marking individual responses.
Can I get my Duolingo English Test rescored?
Duolingo does not offer a rescoring service in the way IELTS and TOEFL do. If you believe a result is wrong, the practical route is to sit the test again, subject to the limit of three purchases in any 30-day period.
Does the DET have a pass mark?
No. It reports a score on a 10-160 scale and the institution you are applying to decides what counts as sufficient. There is no pass or fail on the test itself.
Why is my score reported in increments of five?
Because the test does not claim to distinguish more finely than that. Reporting a 117 would imply a precision the measurement does not support, so the scale reports in fives — a deliberate statement about accuracy rather than a rounding convenience.
What is the average Duolingo English Test score?
Roughly 105 to 110, which sits in the B2 band on the CEFR scale. That is an average across test-takers worldwide, so treat it as a reference point rather than a target — the score you actually need is set by the institution you are applying to, and competitive programmes often ask for 120 or more. See which universities accept the DET and the scores they ask for.
What is the difference between the four skill subscores and the four integrated subscores?
The four skill subscores — Reading, Writing, Listening and Speaking — each measure one ability. The four integrated subscores each combine two: Literacy is Reading plus Writing, Comprehension is Reading plus Listening, Conversation is Listening plus Speaking, and Production is Writing plus Speaking. All eight appear on your report, and your overall is the mean of the four skill subscores.
How does the Duolingo English Test decide when to stop?
It is adaptive, so it does not run a fixed number of questions. Each answer updates the engine's estimate of your ability, and a section ends once that estimate is precise enough — measured by the standard error around it — rather than after a set count. That is why two people can answer different numbers of questions and both get a reliable score.
Start a practice test
An hour under real conditions, all eight subscores, and feedback on every written and spoken answer.
Take a test