AWS Certification Pass Rates Are Not Published: How Scoring Actually Works and How to Judge Your Own Readiness

"What is the pass rate for AWS certification?" is one of the most searched questions about AWS exams, and it has an uncomfortable answer: AWS has never published one. Every percentage you find in a blog post or a forum thread is someone's estimate, and none of them trace back to AWS. This article does not add another guess to the pile. Instead, it covers what AWS does document — the passing scores, the 100-1,000 scaled score, compensatory scoring, and unscored questions — and turns that into a concrete way to decide whether you are ready to sit the exam.

The short answer

AWS does not publish pass rates for any of its certification exams. Every "pass rate is about XX%" figure online is an estimate with no official basis. What is official is the passing score: 700 for CLF-C02, 720 for SAA-C03, and 750 for SAP-C02 on a 100-1,000 scale. Judge your readiness against those, not against a rumored percentage.

Passing Scores We Can Verify Officially

AWS certifications are organized into four tiers — Foundational, Associate, Professional, and Specialty — and the passing score is set per exam, not per tier. The table below lists only the exams whose passing scores NicheeLab has verified directly in the current official exam guides. For every other exam, we deliberately leave the cell blank rather than repeat a number we have not confirmed: check that exam's own official guide.

Notice the pattern: the bar rises as the tier does. Going from CLF to SAA adds 20 points, and from SAA to SAP another 30. That is a meaningful jump, because a scaled score is not a raw percentage — 30 points at the Professional level represents a real gap in how consistently you must reason about unfamiliar scenarios. If you want the difficulty side of this story rather than the scoring side, see AWS certification difficulty compared.

How AWS Scoring Actually Works

The reason a pass rate feels so necessary is that most candidates do not know how the score is produced. Once you understand the mechanics, the need for a pass rate largely disappears. There are four things worth knowing.

• Scaled scoring: Your score is reported on a 100-1,000 scale. It is not your raw correct count expressed as a percentage; it is adjusted so that different versions of the same exam demand equivalent ability. Two candidates who both score 720 may have answered a different number of items correctly.
• No per-domain cutoff: Pass/fail hinges on the overall score alone. The per-domain feedback on your score report is there to guide further study, not to gate the result. A weak domain can be offset by strength elsewhere.
• Unscored questions are mixed in: A fixed number of items are included to evaluate future exam content and do not affect your score. You cannot tell which ones they are during the exam.
• Unanswered means wrong; there is no guessing penalty: Leaving an item blank is scored the same as getting it wrong. Because there is no deduction for a wrong answer, always selecting something is strictly the better strategy.

How Many Questions Actually Count

The unscored-item mechanic surprises people, so it is worth putting the numbers side by side. The exam presents more questions than it grades, and the gap is not small — roughly a fifth of what you see on CLF and SAA does not count.

Two practical consequences follow. First, a strange question is not evidence that you are failing — it may simply not count. Candidates who spiral after three unfamiliar items in a row lose far more points to panic than to the items themselves. Second, the per-question time budget is tighter than it looks, because the clock runs on all questions including the unscored ones. On CLF you have well under a minute and a half per item; treat anything you cannot resolve in that window as a flag-and-move-on.

Why Pass-Rate Numbers Circulate Anyway

If AWS never published a number, where do all the percentages come from? The mechanism is worth spelling out, because it explains why the figures are not just unofficial but structurally wrong.

• The sample is not the candidate population: The tallies come from social posts and blog write-ups — that is, from people who both took the exam and chose to publish the outcome. That is not a cross-section of all candidates; it is a self-selected sample.
• Passes get posted more than failures: People publish wins more readily than losses. That asymmetry pushes the tally upward. Articles built only from failure stories skew the other way.
• Numbers harden through citation chains: Once a figure appears somewhere, the next article cites it without checking the source, and the one after that inherits it. After three hops it has become 'the number everyone sees', and by the fourth it wears the face of established fact. Trace it back and there is usually nothing there.
• Exam versions get mixed together: Exams get revised. When reports from an older version and the current one land in the same tally, the resulting figure does not even define which exam it describes.

The deeper problem is that a pass rate would not help you even if it were accurate. A 70% pass rate does not mean you have a 70% chance — your outcome is determined by your own preparation, and nothing about the aggregate transfers to the individual. Reading "the pass rate is high" as permission to book the exam early is how people end up in the post-failure fourteen-day wait.

Self-Assessment Metrics to Use Instead

Here is what to replace the pass rate with. These thresholds are NicheeLab's operational recommendations, not official AWS guidance; they are set above the nominal passing line on purpose, because a mock exam at home is easier than the real thing under a clock in a test center.

The last row is the one candidates skip, and it is the most diagnostic. Track, per question, whether you chose the answer because you could explain why the other options were wrong, or because it "felt right". A score of 85% where a quarter of the correct answers were coin flips is really a 65% — and a 65% does not survive a different question set. Because there is no guessing penalty you should still answer everything on exam day, but during preparation, an unexplained correct answer should be logged as a miss.

What 'Ready to Pass' Looks Like, by Tier

Thresholds are one thing; knowing what the state actually feels like is another. Here is the qualitative version for each tier.

• Foundational (CLF-C02): You can say in one sentence what any given service does. You do not need configuration details, but you must be able to distinguish S3 from EBS from EFS without hesitating, state where the shared responsibility boundary sits, and explain the pricing models (On-Demand, Reserved, Savings Plans, Spot). If you have memorized service names but cannot explain what they are for, you are not there yet.
• Associate (SAA-C03): You can pull the constraints out of a requirements paragraph and narrow four options down to two by elimination. SAA questions do not ask which design works but which design best meets the requirements, so two of the four options usually do work. The dividing line is whether you can tell when availability is the constraint and when cost is.
• Professional (SAP-C02): You can compare whole multi-service architectures on their trade-offs. SAP answer options are long and every one of them looks plausible. The themes — multi-account design, migration strategy, incremental improvement of an existing estate — cannot be settled with single-service knowledge. Stacking up Associate-level facts does not get you there.

What Our Own Question Bank Shows

Since no official pass rate exists, one honest substitute is to show how the difficulty of the material itself scales across tiers. NicheeLab builds and tags its own AWS question bank, and the difficulty distribution across the 1,200 practice questions for these three exams is a measured figure, not an impression.

The hard share roughly doubles from CLF to SAA and then nearly triples again into SAP, while the easy share collapses from 149 questions to 31. That last number is the one to sit with: at Professional level there is almost no material that can be answered on recall alone. This is the concrete shape behind the 30-point gap between the 720 and 750 passing lines. For a fuller treatment of tier-to-tier difficulty, see the difficulty comparison; for whether the investment pays off at all, see is AWS certification worth it.

A Procedure for Deciding Your Exam Date

Putting it together, here is the sequence we recommend in place of looking for a pass rate.

• 1. Open the official exam guide for your exact exam code and read the passing score and the domain weightings. This is the only primary source.
• 2. Order your study by domain weighting. Because there is no per-domain cutoff, weighting your time toward the heaviest domains raises your expected overall score.
• 3. Close your gaps with domain drills, then take a fresh mock exam under the real time limit. An untimed mock tells you nothing about readiness.
• 4. Repeat until you meet the self-assessment thresholds above — and always track the share of correct answers you cannot explain.
• 5. Book once you clear the bar. Note that a failure carries a 14-day wait before you can retake, and each attempt is charged in full — sitting the exam below the bar costs you both time and money.

If you are working toward a specific exam, the per-exam articles go deeper: AWS Cloud Practitioner overview, SAA difficulty, the SAA pass-rate question, and the SAP-C02 overview. For how long all of this takes, see AWS certification study time by tier.

Frequently Asked Questions

Exam specifications in this article are based on the official AWS exam guides for CLF-C02, SAA-C03, and SAP-C02, together with the AWS Certification testing policies and the AWS Certification FAQ. Difficulty distributions are measured from NicheeLab's own AWS question bank. Self-assessment thresholds are NicheeLab recommendations, not official AWS criteria. This article is not an official AWS product and has no partnership or sponsorship relationship with Amazon Web Services. AWS and Amazon Web Services are trademarks of Amazon.com, Inc. or its affiliates. Information reflects official materials as of August 7, 2026; always confirm the latest details on the official pages.