The Databricks Certified Machine Learning Associate exam costs $200 USD and gives you 90 minutes for 48 scored multiple-choice or multiple-selection questions, delivered online with a proctor. There are no prerequisites and no test aids, and the certification is valid for two years.
Last updated August 2026.
| ML Associate | |
|---|---|
| Registration fee | $200 USD (plus local tax) |
| Scored questions | 48 |
| Unscored items | May also appear, not identified |
| Time limit | 90 minutes |
| Question types | Multiple choice or multiple selection |
| Delivery | Online proctored, no test aids |
| Prerequisites | None (6 months hands-on recommended) |
| Passing score | Not published by Databricks |
| Validity | 2 years, full re-exam to recertify |
Figures from the official exam guide PDF (1 Mar 2025 edition) linked from the Databricks Machine Learning Associate certification page.
Two formats appear: multiple choice, where one option is correct, and multiple selection, where you pick several. Multiple-selection items are the ones that punish partial knowledge, because recognising three correct answers out of four does not earn a partial score.
Code shows up throughout, and the language split is fixed by the exam guide: Python for all machine learning code, with SQL appearing only for non-ML data manipulation. If you have been doing your feature work in SQL, practice reading the equivalent in Python and Spark ML before test day.
The guide lists exactly 48 objectives and the exam has exactly 48 scored questions, so coverage works out at roughly one question per objective. That makes the objective list an unusually literal study checklist: anything on it is fair game, and nothing off it is.
The exam guide is specific here, and it is a detail most summaries get wrong. 48 is the number of scored questions. Unscored items may also be presented, they are not identified when you sit the exam, they do not affect your score, and extra time is factored into the 90 minutes to account for them.
Practically, this means two things: the question count on screen may exceed 48, and you should not spend the last ten minutes agonising over one item on the theory that every question counts. Some of them do not.
There is no official cut score for this exam. Databricks sets pass marks through statistical analysis and adjusts them as exam versions change, which is why no percentage appears in the guide. Any "you need 70%" figure you find quoted online is not sourced from Databricks.
That is not a reason to study without a target. We use 70% on practice exams as our own readiness bar, clearly labelled as our guidance rather than an official threshold: it leaves room for the difficulty gap between practice questions and the real thing.
Retakes are booked as a new registration, so budget the full fee again unless you hold a voucher. Confirm the current retake policy on the Databricks certification site before rebooking.
The registration fee is $200 USD, plus local tax in some countries. A retake is booked as a new registration at the same fee unless you have a voucher.
There are 48 scored questions and a 90-minute time limit. Unscored items may also appear, are not identified during the exam, and do not affect your score, with extra time factored in for them.
Databricks does not publish a passing score for this exam. Cut scores are set by statistical analysis and change as exam versions are updated, so treat any specific percentage quoted elsewhere as unsourced.
Python. The exam guide specifies Python for all machine learning code, with SQL appearing only for non-ML data manipulation. Practising the Python and Spark ML equivalents of your usual SQL feature work is worth the time.
Yes. Delivery is online proctored, so you sit it from home or a private office. You need a quiet private room, a webcam and microphone on for the whole session, photo ID, and no test aids.
Thirty short chapters mapped to the four exam sections, then five full practice exams timed like the real thing.
Start Chapter 1 →