How Multilevel Speaking Is Scored

Every Multilevel speaking attempt is reduced to one number: a score out of 75. That number is what maps to your CEFR band, and it's built from a holistic assessment of each of the four parts — not a checklist of grammar points ticked off one by one.

The official DTM/UZBMB mapping splits the scale into five bands:

• C1 65–75 — fluent, precise, near-native control.
• B2 51–64 — confident and spontaneous, occasional slips.
• B1 38–50 — gets the message across, with gaps.
• A2 20–37 — simple exchanges, frequent hesitation.
• A1 0–19 — isolated words and formulaic phrases.

Full descriptors for each band are on the CEFR bands page.

Four dimensions, applied per part: grammatical range and accuracy, vocabulary, fluency, and coherence and cohesion. Pronunciation isn't scored as a separate criterion — it's not something a rubric judges from a transcript, so it doesn't move the holistic band the way the other four do.

That's worth internalizing if you've been told your accent is the problem: on this exam, it almost never is. Structure, range, and getting through a full turn without losing the thread matter far more.

Under the hood, each part is first scored on its own raw scale, then combined. Part 1.1, Part 1.2, and Part 2 are each marked out of 5; Part 3 is marked out of 6. That 21-point raw total is then converted to the headline 0–75 score using the official conversion table. So Part 3 — already the hardest part on paper, targeting C1 — also carries the single biggest share of the raw total. See the exact split on the score calculator.

The four parts step up in difficulty by design — Part 1.1 targets A2, Part 1.2 targets B1, Part 2 targets B2, Part 3 targets C1. A test-taker who nails Part 1.1 but stalls on Part 3's argument discussion isn't failing randomly — they're hitting the edge of their actual range. Knowing which part that is changes how you should spend your practice time; see the single part practice guide for how to drill it directly.