What a TCF score is, and why nobody can calculate one from your answers
Search for a TCF score calculator and you will find a dozen. Type in how many questions you got right and one will hand you a number like 487. That number cannot be right, and the people publishing it know it cannot. Here is what is actually published, what is not, and what we do instead.
What the TCF publishes about its own scoring
France Éducation international, which owns and runs the TCF, describes the conversion in its Manuel du candidat in one sentence:
Un barème est appliqué aux bonnes réponses. Il prend en compte la difficulté de chaque question et transforme le nombre de bonnes réponses obtenues en un score de 100 à 699.
And, elsewhere in the same manual, that the overall score “is the result of a calculation taking into account the number and the degree of difficulty of the questions, and is not a simple arithmetic average.”
So the barème exists, and the exam board says it weighs difficulty. It has never published it. Not the table, not the formula, not the model, not one worked example — in any version of the candidate manual we could find, from 2014 to April 2025.
And a fixed table could not be correct even if they published one
The same manual states that “whatever the version of the test, results remain comparable.” That is a technical claim, and it has a consequence most calculators quietly ignore.
Different sittings use different questions, and some sets are harder than others. To keep results comparable, the conversion has to differ per version — an easier paper needs more correct answers for the same score. That means no single fixed table of “28 correct = 458” can be right for every sitting. A calculator that gives you one number regardless of which paper you sat is not simplifying. It is guessing.
Two candidates with the same number of correct answers, on different versions, can hold different official scores. That is the system working as designed.
What is published, and we use it
One link in the chain is genuinely official and public: the conversion from a TCF score to an NCLC level. Immigration, Refugees and Citizenship Canada publishes it, because candidates have to be able to read their own results.
Listening — NCLC 4: 331–368 · 5: 369–397 · 6: 398–457 · 7: 458–502 · 8: 503–522 · 9: 523–548 · 10+: 549–699
Reading — NCLC 4: 342–374 · 5: 375–405 · 6: 406–452 · 7: 453–498 · 8: 499–523 · 9: 524–548 · 10+: 549–699
Speaking and writing, out of 20 — NCLC 4: 4–5 · 5: 6 · 6: 7–9 · 7: 10–11 · 8: 12–13 · 9: 14–15 · 10+: 16–20
Notice the shape of that last row. NCLC 7 spans two marks. NCLC 5 spans exactly one. The bands are not evenly sized, which is why any product claiming a single accuracy figure across all levels is telling you less than it appears to.
So: if you have an official score, converting it to NCLC is exact and we do it. Going the other way — from answers to a score — is the part nobody outside the exam board can do.
What we do instead
Our practice sections give you three things, and each is exactly what it says it is.
How many you got right. Out of 39, per section. This is exact, and it is ours — our questions, our answer key.
Where you broke down. Every question we write is tagged to a difficulty level from A1 to C2. So the result reads “solid through B1, falling apart at B2” rather than as a three-digit number. That tells you what to study. A score of 487 does not.
Nothing else. We do not print a TCF score, and we do not print a range dressed up as one.
The difficulty banding is ours, not the exam board's. The TCF publishes that its comprehension sections rise in difficulty; it does not publish which question sits at which level. Our banding is an authoring decision and we label it as one — on this page and on the result screen.
How to read a TCF calculator you find elsewhere
Two kinds exist, and only one of them is honest.
Score → NCLC. You type in the score from your real attestation and it gives you a level. This is a lookup of a published table. It is correct, and it is useful.
Answers → score. You type in how many you got right and it gives you a number on the 100–699 scale. There is no published basis for this. If a site offers it, the fair questions are: where did the conversion come from, does it change with the version of the exam, and why is the source not cited?
One widely-copied table on the open web assigns points per question rising from 3 to 33 and says they total 699. It traces back to a preparation blog, not to the exam board — and it cannot be right on its own terms: an additive table totalling 699 means zero correct answers scores zero, and the published floor of the TCF scale is 100.
What would let us do better, and when
There is an honest route to an estimate, and it is slow. Collect real official score reports from candidates who consent to share them, alongside the practice work those same candidates did, and fit our own scale to that evidence. Then publish the error — how far our estimate sits from the real result, stated per level, not as one flattering average.
In build Until that exists, our comprehension sections report a count and a difficulty profile, and no scale score. If we reach it, what we publish will be our error, including where it is worst.
Sources on this page: the Manuel du candidat TCF published by France Éducation international, and the official test equivalency charts published by Immigration, Refugees and Citizenship Canada. We link to primary sources rather than to summaries of them, and where we could not find a primary source we say that instead of filling the gap.