CELPIP Score Chart and CLB: What Is Official, and What Everyone Else Is Guessing

The CELPIP-to-CLB part is simple and official. The part everyone actually searches for — how many correct answers you need — is not published at all.

IRCC prints a plain chart: CELPIP 7 is CLB 7, CELPIP 9 is CLB 9, straight across all four skills. That part is settled. But nearly every "CELPIP score chart" page also prints a second table — the one telling you that some specific number out of 38 is a level 9. Those numbers are reverse-engineered by prep companies, they contradict each other, and CELPIP labels its own version "Approximate".

So this page does the split: here is what is official, here is what is guessed, and how to tell which is which.

The failure

Two mistakes, both of which end with someone planning around a number that does not exist.

The first is averaging. Your score report shows an "Average Score" and it looks like a level. People take it as one, decide they have "a CLB 8", and stop.

The second is trusting a raw-score cutoff. You sit a practice test, look up a chart that turns your Reading raw score into a 9, and conclude you are done studying. A different chart makes the same result an 8. Neither chart is from IRCC, and neither is a rule.

The chart that is official

Each component — Listening, Reading, Writing, Speaking — gets its own CELPIP Level. There is no single overall CELPIP level, which is why the chart below has no "total" row.

The right-hand column is the part other pages flatten.

IRCC's own equivalency chart publishes CELPIP N = CLB N for CLB 4 through 10, identically across Reading, Writing, Listening and Speaking. Those seven rows are the ones that a government page will back you up on.

Paragon — the company that makes CELPIP — states that "CELPIP test scores have been aligned to the Canadian Language Benchmarks (CLB) Levels 3 to 12", and its guidebook chart runs the one-to-one mapping up to 11 and 12 and down to 3. Those rows are real, but if you say "CELPIP 11 equals CLB 11", your source is the test maker, not the department deciding your file.

At the bottom the tidy pattern breaks. CELPIP 2 maps to CLB "1, 2" — one CELPIP level covering two CLB levels. Levels 0 and 1 map to no CLB level at all — they carry "Insufficient information to assess". (The old "M" score was replaced in the charts "with the 0-2 score" in December 2023, so older screenshots showing an "M" are out of date.)

Before you act on any of this, open IRCC's equivalency chart and read your own row. This page is a guide to what the sources say, not immigration advice — the government page is the one that counts.

The "Average Score" trap

Your score report shows an Average Score. It is not a level, and Paragon says so in plain words:

"Average Score is not equivalent to any Canadian Language Benchmark (CLB) level"

Nothing in the CLB system is built out of an average. Programs are written against per-skill minimums, and an average cannot tell you whether you met one — two people with the same average can be in completely different positions if one of them has a single low skill.

If you have been carrying around a single number for your CELPIP result, go back to the report and write down four numbers instead.

Your weakest skill is what gets measured

This follows directly from "each component gets its own level". When a program states a minimum across all four skills, it is your lowest component that decides the outcome, not your average and not your best. A 10 in Listening does not carry a 6 in Writing.

The practical consequence for study time is unglamorous: the hours are worth more in your worst skill, even when your best one is more enjoyable to practise. Which minimum applies to you depends entirely on the program — check IRCC's page for the one you are applying under.

The table this page will not print

You came here for a raw-score cutoff. There is not going to be one, and here is exactly why.

Listening and Reading are each reported against a raw score out of 38, so a mapping from "correct answers" to "CELPIP level" clearly exists somewhere inside Paragon. What does not exist is a published, binding version of it. CELPIP labels its own raw-score table "Approximate". When the organisation that builds the test hedges its own chart, a third-party chart that does not hedge is not being more helpful — it is being less honest.

And those third-party charts disagree. Compare two prep sites and you will find the same Reading raw score sorted into different levels, with neither page citing anything. They are reverse-engineered, not published by Paragon or IRCC. Once you know that, a confident cutoff table stops being reassuring and starts being a warning sign about the rest of the page.

One thing here is official and worth more than any cutoff: "there are no points deducted for wrong answers". So the only strategy the raw score licenses is the simple one — never leave a blank. A guess is free. An empty answer is a guaranteed zero on that question.

What actually costs you

• Treating the Average Score as a CLB level. It is not one, and Paragon says so on its own site.
• Quoting a raw-score cutoff you found on a prep site as though it were a threshold. Nobody publishes one.
• Studying your strongest skill because practice feels good there, while your weakest one is the one being measured.
• Leaving blanks in Listening or Reading to "avoid mistakes". There is no penalty for a wrong answer.
• Quoting CELPIP 11 or 12 as CLB 11 or 12 without naming the source. IRCC's chart stops at 10.
• Reading an old chart with an "M" on it. That notation was replaced in December 2023.

Try one

[Try a marked task →] — one full task, marked, no card required.