Conversion chart

CELPIP Listening Score Chart

Every raw score out of 38, with the CELPIP level the published table gives it and the honest range we report for it. The CELPIP level is also your CLB level. These are estimates from published material, not official scores.

Open the interactive score calculator

The full listening conversion chart

The source band and source level columns are Paragon’s sample-test cut-offs, unchanged. The our estimate column widens that level by one, in the direction your score leans inside its own band, because a raw total does not pin a level down on its own — see how to read this chart below.

Every raw score, 38 down to 0 – Listening
Raw correct	Source band	Source level	Our estimate	Where in the band
38	36–38	10–12	10–12	top band, not narrowed
37	36–38	10–12	10–12	top band, not narrowed
36	36–38	10–12	10–12	top band, not narrowed
35	34–35	9	9–10	upper half of band
34	34–35	9	8–9	lower half of band
33	32–33	8	8–9	upper half of band
32	32–33	8	7–8	lower half of band
31	29–31	7	7–8	upper half of band
30	29–31	7	6–7	midpoint (rounds down)
29	29–31	7	6–7	lower half of band
28	24–28	6	6–7	upper half of band
27	24–28	6	6–7	upper half of band
26	24–28	6	5–6	midpoint (rounds down)
25	24–28	6	5–6	lower half of band
24	24–28	6	5–6	lower half of band
23	19–23	5	5–6	upper half of band
22	19–23	5	5–6	upper half of band
21	19–23	5	4–5	midpoint (rounds down)
20	19–23	5	4–5	lower half of band
19	19–23	5	4–5	lower half of band
18	13–18	4	4–5	upper half of band
17	13–18	4	4–5	upper half of band
16	13–18	4	4–5	upper half of band
15	13–18	4	3–4	lower half of band
14	13–18	4	3–4	lower half of band
13	13–18	4	3–4	lower half of band
12	8–12	3	3–4	upper half of band
11	8–12	3	3–4	upper half of band
10	8–12	3	3 or below	midpoint (rounds down)
9	8–12	3	3 or below	lower half of band
8	8–12	3	3 or below	lower half of band
7	0–7	M	below 3	not reported as a level
6	0–7	M	below 3	not reported as a level
5	0–7	M	below 3	not reported as a level
4	0–7	M	below 3	not reported as a level
3	0–7	M	below 3	not reported as a level
2	0–7	M	below 3	not reported as a level
1	0–7	M	below 3	not reported as a level
0	0–7	M	below 3	not reported as a level

Cut-offs: Paragon Free Online Sample Test, client-side function findListScore. Recorded in docs/LISTENING_FORMAT_NOTES.md §6.

How to read this chart

Find your raw score, not your percentage

The Listening section has 38 scored questions. Count the ones you got right; that whole number is the only input. There is no penalty for a wrong answer in the published tables, so an unanswered question and a wrong one are worth the same — guess rather than leave a blank.

Read the band, then read the range

A band such as 29–31 → level 7 means every raw score in it returns the same level from the source table. Our estimate then adds one level of give: a score in the lower half of its band reports as 6–7, one in the upper half as 7–8. A score exactly at the midpoint rounds down.

Why a range and not a number

The real exam is equated: levels are computed from the number of points and the difficulty of the questions, and scores are made comparable across test forms. The same raw total can therefore land a level apart on two different forms. Paragon’s two published tables also disagree with each other by a level at several raw scores. A chart that answers with one number is hiding that, and results day is a bad time to find out.

M is not a level

A raw 0–7 returns M in the source code. M is a refusal to report a level rather than a low level, which is why this chart shows it as “below 3”.

One part cannot be converted

This chart converts a full 38-question section. A single part on its own has no published conversion, and scaling one up to 38 would be arithmetic dressed as measurement.

Do not use the Reading chart for Listening

The two sections do not share a table, and the difference is large enough to matter: a raw 30 out of 38 is Reading level 8 but Listening level 7, and the top band starts at 34 for Reading and 36 for Listening. Both sets of cut-offs come from the same sample test — calcReadScore and findListScore are two separate functions in it.

The gap is consistent with the official 2024 distribution, where the Listening mean is 8.22 and the Reading mean is 7.21. Candidates score higher on Listening, so a given level takes more correct answers there. A chart that gives one table for both sections is over-scoring your Listening by roughly a level in the middle of the range.

Is there a second Listening table?

Not one we could find. Paragon publishes a printed conversion table in the Reading Pro Study Pack, and we show it on the Reading chart page. We could not locate an equivalent published Listening table in any source we checked, so this page has one table rather than two. Writing a second one from the Reading bands, or from a prep site’s reconstruction, would be inventing data, and we would rather leave the gap visible.

Listening – Paragon sample-test cut-offs
Raw correct / 38	CELPIP level
36–38	10–12
34–35	9
32–33	8
29–31	7
24–28	6
19–23	5
13–18	4
8–12	3
0–7	M

No published Listening counterpart to the Reading Pro Study Pack table was found. This space stays empty until we can cite one.

Paragon Free Online Sample Test, client-side function findListScore.

What the level means in CLB terms

For levels 3 to 12 the CELPIP level is the CLB level, one to one, so this chart is also a raw-score-to-CLB chart. CELPIP 2 falls into the CLB 1–2 band.

CELPIP level, CLB level and CEFR
CELPIP	CLB	CEFR
12	12	C2
11	11	C1
10	10	C1
9	9	B2
8	8	B2
7	7	B2
6	6	B1
5	5	B1
4	4	A2
3	3	A2
2	1–2	A1

Read this before you trust the number

These cut-offs come from Paragon's free sample test, which is an estimator — not the operational scoring engine. The sample test scores itself in your browser. That code is the most authoritative public artefact we could find, but it is not what marks the real exam.
The real exam is equated and difficulty-weighted. Paragon's own Study Pack says the level comes from the number of points and the difficulty of the questions, with scores equated across test forms. So no fixed raw-to-level table can be exact: the same raw total can land a level apart on two different forms.
Paragon's two public artefacts disagree with each other. The sample-test code puts a Reading raw 24 at level 6; the printed Study Pack table puts it at level 7. Both tables are shown on this page rather than blended into a single fake answer.
This is an estimate, not an official score. The official raw-score conversion is not published, and we do not issue official scores. Only Paragon can tell you your CELPIP level.
A single part cannot be converted. These tables convert a full 38-question section. There is no published conversion for an 11-question part on its own, and we will not invent one.

Questions people actually ask

Is this the official CELPIP score conversion?

No. Paragon does not publish the conversion the operational exam uses. The cut-offs here were read out of the free sample test’s own client-side scoring function, which is the most authoritative public artefact we could find, but it is an estimator rather than the engine that marks the real exam. Treat the level as an estimate; only Paragon can tell you your CELPIP level.

What raw score do I need for CLB 9 in Listening?

In the sample-test cut-offs a raw 34 or 35 out of 38 returns level 9, and for levels 3 to 12 the CELPIP level is the CLB level. This page reports it as a range rather than a single number, because the real exam is equated: the same raw total can land a level apart on two different forms.

Can I use the Reading chart for my Listening score?

No, and the gap is large enough to change an application. A raw 30 out of 38 is Reading level 8 but Listening level 7, and the top band starts at 34 for Reading and 36 for Listening. The two sections do not share a table even though both have 38 questions.

Why does this page have one table when the Reading chart has two?

Because we could only find one. Paragon publishes a second, printed conversion table in the Reading Pro Study Pack, and the Reading chart shows it. We could not locate a published Listening counterpart in any source we checked. Building one from the Reading bands, or from a prep site’s reconstruction, would be inventing data, so the gap is left visible instead.

Is my CELPIP level the same as my CLB level?

For levels 3 to 12, yes, one to one — CELPIP 8 is CLB 8. The one exception at the bottom is CELPIP 2, which falls inside the CLB 1–2 band rather than mapping to a single number.

Does leaving a question blank cost more than answering it wrong?

No. The published table applies no penalty for a wrong answer, so a blank and a wrong answer are worth the same — nothing. Guess rather than leave a blank.

Do it without counting rows

The calculator does the same lookup, with both tables beside it and the same caveats attached.

CELPIP score calculator Reading score chart

No raw score to look up yet? A free full form marks itself the moment you finish — CELPIP Listening practice test.

Where these numbers come from

Cut-off thresholds: Paragon Testing Enterprises' free CELPIP-General sample test, read from the product's own client-side scoring functions (calcReadScore for Reading, findListScore for Listening). Recorded in our research notes as docs/READING_FORMAT_NOTES.md §5a and docs/LISTENING_FORMAT_NOTES.md §6.
Listening printed bands: not found. We could not locate a published Study Pack conversion table for Listening in any source we checked, so no second Listening table is shown here. We would rather show one table than invent a second.
CELPIP / CLB / CEFR mapping and the 2024 score distribution: the official CELPIP Guidebook for Test Takers and the 2024 CELPIP Data Report.