Celpip AI Try a scored task

Exam format

CELPIP test format

Four sections, one sitting, about three hours. Below is what each section actually contains — counts, clocks and the rules that decide as much of a score as the questions do. Where the published sources do not say, this page says so instead of filling the gap.

The four sections at a glance

CELPIP-General sections, parts, questions and time
Section Parts Questions Time
Listening6 parts3846–55 min
Reading4 parts3843–56 min
Writing2 tasks53 min
Speaking8 tasksabout 15 min

The ranges are not vagueness. Both Listening and Reading can carry extra questions that are not scored, mixed in and impossible to tell apart, and their minutes are inside the total. The scored content is 43 minutes of Reading and about 36 minutes of Listening; the rest is instructions, scene cards and any trial material.

Listening — 6 parts, 38 questions

CELPIP Listening parts, questions and answer windows
Part Questions Question is Answer window
1 — Listening to Problem Solving8spoken30 s each
2 — Listening to a Daily Life Conversation5spoken30 s each
3 — Listening for Information6spoken30 s each
4 — Listening to a News Item5on screen205 s for all
5 — Listening to a Discussion8on screen240 s for all
6 — Listening for Viewpoints6on screen260 s for all
  • Every clip plays once and there is no playbar. No pause, no rewind, no second listen.
  • In Parts 1 to 3 the question is not written on the screen. You hear it, once, and only the four options are visible. That is a memory load the later parts do not have.
  • Part 1 comes in three clips with a 10-second preparation gap between them; Part 5 is a video rather than audio.

Three full Listening forms run in the browser under these rules, free and unmetered.

Reading — 4 parts, 38 questions, 43 minutes

CELPIP Reading parts, questions and minutes
Part Questions Minutes
1 — Reading Correspondence1111
2 — Reading to Apply a Diagram89
3 — Reading for Information910
4 — Reading for Viewpoints1013
  • Each part is one screen: the passage on the left, the questions on the right, both scrolling on their own.
  • The clock is per part and does not carry over. Finishing Part 1 early does not buy time in Part 2; when a part’s time is up the exam moves on by itself.
  • Part 3 has a genuine fifth option for information that appears in none of the paragraphs. It is a real key, not a courtesy.

Three full Reading forms run in the browser with the same clocks.

Writing and Speaking

Writing — 2 tasks, 53 minutes

Task 1 is an email; Task 2 is a response to survey questions. The exam interface gives you a word counter and a spell checker.

The published source gives 53 minutes for the section as a whole and does not split it between the two tasks. Our simulator uses 27 and 26 minutes, which is our choice, not a published fact.

Both Writing tasks and their whole question pool

Speaking — 8 tasks, about 15 minutes

Giving Advice · Personal Experience · Describing a Scene · Making Predictions · Comparing and Persuading · Difficult Situation · Expressing Opinions · Unusual Situation.

30 or 60 seconds of preparation, then 60 or 90 seconds of speech. No pause button, no second take, and no interviewer — you speak to a microphone. Task 5 adds a separate 60 seconds to choose between its two options.

All eight Speaking tasks and their question pools

The rules that cost people marks

  • You cannot go back. Inside a part you may change an answer until the clock runs out; once you leave the page, it is closed. In Listening Parts 1 to 3 a page is a single question, so those close one at a time.
  • No flagging, no highlighting, no question navigator. The whole control set is Next and Back.
  • The timer only turns urgent at the end. Above a minute it shows whole minutes; under 60 seconds it counts seconds and turns red. There is no other warning.
  • A blank costs the same as a wrong answer. Neither conversion table penalises a guess.
  • An official result is issued 2 to 4 business days after the test.

What the sources do not say

The raw-score conversion the operational exam uses is not published. Two published artefacts exist and they disagree with each other at several raw scores; both are laid out side by side on the Reading and Listening score charts.

The order of the parts can differ. The published order is the one above, but the sample test warns that a real sitting may present question types in another order.

How much unscored trial material appears, and where. Only that it can be there, that it is indistinguishable, and that its minutes are inside the totals.

Everything above is drawn from the test provider’s own published guidebook and free sample test. We are an independent practice tool, not the exam, and this page is a summary of theirs rather than a substitute for it.