How Is CELPIP Scored? Raters, Raw Scores & CLB Levels

Every CELPIP skill lands on a level from M to 12, and those levels are what IRCC reads. How you get there differs by component: Listening and Reading are marked entirely by computer, while Speaking and Writing go to teams of trained human raters, with an AI system assisting on Writing. No examiner sits across from you, and no raw score ever appears on your report.

All Listening and Reading questions are multiple choice or similar, scored dichotomously: correct or incorrect, nothing in between. Questions left blank are marked incorrect, and there's no deduction for a wrong answer. Guessing can only help you.

Your raw count of correct answers isn't what appears on the report. Why not? Test forms differ slightly in difficulty, so CELPIP converts raw scores into scaled scores through a process called equating: 30 correct on a harder form and 30 correct on an easier one produce different scaled results, which keeps the final levels fair. That's also why the same level can need a slightly different number of correct answers from one sitting to the next. For the approximate ranges per level, see our CELPIP score chart.

One more wrinkle: every test contains a few unscored questions being trialled for future exams. They look identical to scored ones, so treat every question as if it counts.

Each speaking performance is assessed by three to five raters, each writing performance by four to six, and they work independently, blind to each other's ratings. If the ratings disagree, a benchmark rater with a strong accuracy record re-assesses the performance. Writing adds one more layer: an AI-human hybrid system scores it first, and the human rating panel reviews and verifies the result.

Raters score four dimensions per skill. Speaking is judged on content and coherence, vocabulary, listenability (rhythm, pronunciation, and grammar in speech), and task fulfillment. Writing swaps listenability for readability: paragraphing, transitions, grammar, and spelling. The full descriptors live in the official performance standards, and our speaking score chart breaks down what each level looks like in practice.

On Celpify, practice writing and speaking attempts are scored against these same four dimensions, and one pattern repeats: task fulfillment drags levels down more often than grammar does. Answering half the prompt beautifully still caps the score.

Scaled scores and rater judgments are transformed into CELPIP levels through standard setting, a research process where language experts define what performance each level requires. The result maps one to one onto the Canadian Language Benchmarks: CELPIP 9 means CLB 9. Our CLB converter handles the cross-test comparisons if you're weighing scores against IELTS requirements.

Two report details trip people up. The Average Score printed on your report isn't used by IRCC at all; immigration programs read each skill separately, so a 7 in writing matters even if your average is 9. And what counts as a good level depends entirely on your program, which is the subject of our guide to what a good CELPIP score is. Scores post to your account in 2 to 4 business days; the results timeline guide covers that clock.