Skip to content

Databricks Data Engineer Professional: exam format and domain weights

The question count, time limit, price and nine weighted domains of the Databricks Certified Data Engineer Professional exam under the 9 October 2026 guide, and what the vendor does not publish.

Published 20 November 2026 · 6 min read

What this guide covers

This page summarises the format of the Databricks Certified Data Engineer Professional exam as it stands under the exam guide published on 9 October 2026. Every figure below comes from CertExam's certification registry, which we verified against the vendor's published guide and certification page on 20 November 2026. If you are about to book the exam, treat the vendor's own pages as the final word and check them again, because vendors revise guides.

CertExam is an independent provider and is not affiliated with or endorsed by Databricks. Databricks is a trademark of its owner and is used here only to describe the exam.

The format at a glance

The exam is a single sitting of multiple-choice questions. The vendor guide describes 60 scored questions and a 120-minute limit. It also says the exam may include unscored questions, up to ten, that are not identified to you and are used to gather statistics. Extra time for these is already built into the 120 minutes, so you should answer every question as if it counts.

  • Scored questions: 60.
  • Unscored questions: up to 10 may appear, unidentified, and do not affect your score.
  • Time limit: 120 minutes.
  • Question type: multiple-choice. The guide does not say whether any items are multi-select, so read each stem carefully for how many answers it asks for.
  • Delivery: online proctored or at a test centre.
  • Language: English.
  • Test aids: none allowed.
  • Price: 200 US dollars.
  • Validity: two years.
  • Prerequisites: none required, though related training is highly recommended.
  • Recommended experience: about one year of hands-on work with the data engineering tasks in the exam outline.

The nine domains and their weights

The guide splits the exam into nine sections, and the weights add up to 100 percent. They are the single most useful planning tool you have, because they tell you how many of the 60 scored questions to expect from each area on average.

  • Developing Code for Data Processing using Python and SQL: 23 percent.
  • Cost & Performance Optimization: 15 percent.
  • Data Ingestion & Acquisition: 12 percent.
  • Data Manipulation: 12 percent.
  • Monitoring and Alerting: 10 percent.
  • Debugging and Deploying: 10 percent.
  • Ensuring Data Security and Compliance: 8 percent.
  • Data Governance: 5 percent.
  • Data Modeling: 5 percent.

What the weights mean for your time

As a rough guide, 23 percent of 60 scored questions is about 14 questions from the first domain, while the two smallest domains contribute about three questions each. That does not mean the small domains can be skipped. Three questions is three questions, and a domain with few objectives can be learned quickly, so it is often the cheapest place to gain marks.

It does mean the largest domain deserves the largest share of your study hours. It is also the broadest: the guide lists twelve objectives under it, covering Python project structure for bundles, dependency troubleshooting, user-defined functions, streaming pipelines, change data capture, Lakeflow Jobs control flow, compute choices, testing, and Structured Streaming state handling. The guide's objectives are the real syllabus, and our study-plan guide breaks them down week by week.

Pacing follows from the arithmetic. With 60 scored questions in 120 minutes you have an average of two minutes per question, a little less if unscored items are present. Timed mock exams are the way to find out whether your reading speed fits that average, and they are most useful once your content knowledge is in decent shape.

What changed from the previous version

The exam was revised on 9 October 2026. According to the vendor's transition guide, the previous version ran through 8 October 2026 with 59 questions, ten sections and four languages. The current version has 60 questions, nine sections and English only. Material on data sharing and federation was folded into the ingestion section, and topics such as Iceberg, the VARIANT data type, AI functions, change data capture through Lakeflow Connect, Clean Rooms and metric views were added.

The vendor also renamed several features, and the current guide uses the new names. If you are studying from older material, expect some of the vocabulary to differ. Check any resource you use against the current guide before trusting its terminology.

What the vendor does not publish

Databricks does not publish a passing score for this exam. Its certification page says the passing score is not published, is set through statistical analysis and can change as questions are added. You will sometimes see a percentage quoted on forums or in other providers' marketing. Without a vendor source it is a guess, and CertExam never states a pass mark that the vendor has not published.

No exam code is published for this exam in the sources we checked.

One more caution. When we verified the registry, the vendor's certification page body still showed some older details, such as 59 questions and more languages, while linking to the new guide. We treated the guide as authoritative. If the two disagree when you read them, the guide's date is the thing to check first.

Where to go next

Read the vendor's exam guide itself first: it is free, and the objectives are written there in the vendor's own words. Then decide how you will cover each domain. Our six-week study plan turns the weights above into a schedule, and the guide on practice questions explains how to use them without fooling yourself.

The certification page for this exam says whether CertExam offers a practice bank for it, how many questions have been independently validated and what the bank's status means.

Keep reading