The AWS SAA Pass Rate Is Not Published: What "XX%" Really Is, and What to Measure Instead

Search for the AWS SAA pass rate and you will find confident percentages everywhere. Follow any of them back to a source and the trail goes cold. That is not an accident of sloppy writing — it is a direct consequence of a fact that most of those articles never mention: AWS does not publish pass rates. This article treats that absence as the story. We lay out precisely what AWS does publish about SAA-C03 scoring, explain what the missing number would and would not have told you, and replace it with something you can actually act on: measurable thresholds from your own practice results.

The short answer

AWS does not publish a pass rate for SAA-C03. Every "about XX%" you have read is an estimate with no official source. What is published is the passing score — 720 out of 1,000 — so measure yourself against that, not against a rumor.

What Is Published and What Is Not

The cleanest way to see the problem is to separate the two categories. Everything in the left column below comes from AWS's own exam guide and policy pages. Everything in the right column does not exist in any official source — which means anyone quoting it is quoting something else.

There is one wrinkle worth stating honestly rather than smoothing over. The AWS Certification FAQ contains a statement to the effect that AWS does not publish passing scores, and that a passing score may be updated to reflect changes in exam format that come with content updates. Meanwhile the SAA-C03 exam guide states 720 explicitly. The two are not perfectly consistent, and the FAQ wording may simply predate the current guides. We are not going to resolve that for you — but note what it is not evidence of: the FAQ is discussing the passing score, not the pass rate. Citing it as "AWS says it doesn't publish pass rates" would be a misquote.

Why a Pass Rate Would Be Hard to Interpret Anyway

Suppose AWS did publish one. What would it mean? The denominator would be everyone who sat SAA-C03 in some window — a group that mixes first-time candidates, people on their third attempt, engineers with five years of production AWS experience, and students taking it as a course requirement. Change the marketing push, change the corporate training programs feeding candidates in, change the mix of countries, and the number moves without the exam having changed at all.

There is a second reason specific to how AWS scores. The passing standard is set against statistically calibrated item difficulty, which is what scaled scoring is for: two candidates who see different question sets of unequal difficulty should face the same standard. A well-calibrated exam therefore holds the standard constant while the observed pass rate drifts with the population. The number you would be reading is mostly a measurement of who showed up. That is precisely why it is a poor input for the decision you are actually trying to make, which is "am I ready to book the exam?"

Self-Assessment Thresholds to Use Instead

Here is the replacement. The table below is NicheeLab's own guidance, not an official AWS standard — we are stating that plainly because the entire point of this article is that unsourced numbers should be labeled. These thresholds come from how our AWS question sets are built and graded across 2,225 questions, and they are framed as a booking decision: hold, book, or comfortable.

Two of these rows do more work than the others and are worth dwelling on.Stability matters more than any single high score: one strong result on one set tells you about the set, not about you. Three sets within a 10-point band is evidence that your knowledge, rather than the sampling, is producing the number.Domain balance matters because SAA-C03 has no per-domain cut score — which sounds like it should make balance irrelevant, and does the opposite. With no cut score, your total is a weighted sum, so a domain sitting at 55% quietly caps your ceiling no matter how strong the rest is. The free 20-question official set is listed first because it is free and calibrated to official phrasing, but 20 items is too small a sample to carry a booking decision on its own — treat it as a phrasing check, not as a score.

One more calibration point from our own data. In our SAA-C03 practice pool of 400 questions, difficulty is distributed easy 122 / medium 231 / hard 47. Medium items are about 58% of the pool, and that is where pass-or-fail is decided. If your practice accuracy is high on easy items and collapses in the medium band, a higher overall percentage from an easy-weighted set will flatter you. Check the shape of your accuracy, not just its average — this is the same reason we recommend judging a question bank by its difficulty mix rather than its question count.

Score Standards Across the Three Exams

None of the three AWS certifications in this tier publishes a pass rate, but all three publish a passing score — and the differences between them are informative in a way a pass rate would not be. The passing score climbs with tier, and so does the share of the exam that is scored.

Read the gap between 700, 720, and 750 as the tier signal it is: the standard tightens by 20 points from Foundational to Associate and by another 30 from Associate to Professional. Our own question data shows the same gradient from a different angle — in our 400-question practice pools, hard items make up roughly 7% for CLF-C02, 12% for SAA-C03, and 31% for SAP-C02. Note also the yen prices: AWS states in its FAQ that yen-denominated exam fees are updated annually in April, and whether the listed price includes tax is not stated explicitly on the official page, so treat the figures above as the listed price rather than a final charge. Full context on how the three compare is in AWS certification pass rates.

You Cannot Compute How Many You Can Miss

This is the question the pass-rate search is usually a proxy for: how many can I get wrong and still pass?The answer is that no one can tell you, and the reason is structural rather than secretive.

• Only 50 questions are scored: of the 65 questions presented in SAA-C03, only 50 count toward your result. The other 15 are unscored items used to gather statistics for future exams.
• You are never told which 15: neither their position nor any way to spot them is disclosed. There is no viable "this one probably doesn't count" strategy — you have to answer all 65 in earnest.
• The score is not a percentage: what you see is a scaled score from 100 to 1,000, statistically adjusted for item difficulty. No raw-to-scaled table is published.
• Blanks are scored as wrong: there is no penalty for guessing, so selecting an answer on every question you are unsure about is always the higher-expected-value move.
• Only two question formats: multiple choice and multiple response. Multiple-response items state how many to select, so you do not need a time strategy built around partial credit.

The practical translation of all five points is a single rule: build your margin in practice, not in the exam room. Because you cannot know which items count, the only lever you control is raising your accuracy across the whole set. A candidate consistently at 80% on unseen 65-question mocks has enough margin that the identity of the 15 unscored items stops mattering. That is the entire practical replacement for a pass rate. How to get there is covered in AWS SAA study methods, and what to do if the first attempt does not go your way is in what to do after failing AWS SAA.

How to Read an "XX% Pass Rate" Claim

You will keep encountering these numbers, so here is a short filter. Ask three questions of any pass-rate claim you find.

• What is the denominator?: all candidates worldwide, one school's students, or respondents to a social-media poll? A pass rate with no stated denominator is not a rate at all.
• When is it from?: figures from earlier versions of the exam circulate with only the exam code swapped out. When the exam is revised, the basis changes.
• Can you follow the source?: in most cases the link goes to another roundup article, which links to another roundup. A citation chain that never reaches AWS is not evidence.

Note that this filter is not a claim that all such numbers are wrong. Some may well be close to reality. The point is that you have no way to tell which, and a number you cannot verify is not a number you can plan around. Meanwhile the passing score, the question count, the scored/unscored split, the time limit, and the retake policy are all published, verifiable, and directly actionable. Build your decision out of those.

Frequently Asked Questions

Exam specifications in this article are based on the official exam guides for CLF-C02, SAA-C03, and SAP-C02, together with the AWS Certification FAQ and the before-testing policies. AWS does not publish pass rates; no figure in this article is presented as one. The difficulty distributions are NicheeLab's own measurements from our AWS question set. 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.