AIGP Passing Score: Why 300 Doesn't Mean 60%

Every AIGP candidate eventually asks the same question: how many questions do I actually need to get right? It sounds simple. The answer is more nuanced than most prep guides admit — and understanding it properly changes how you study, since studying for the wrong target number is one of the more avoidable reasons candidates come in under-prepared.

Let's break down the scoring system in full, because knowing the mechanics of the benchmark is itself a strategic advantage.

The Definitive Passing Score

The AIGP uses a scaled score, not a raw percentage. The passing benchmark is a scaled score of 300, on a range from 100 to 500. This number does not fluctuate between candidates or exam versions. It is the single, fixed standard you must reach, regardless of which exam form you receive on test day.

A scaled score of 300 is not "60% correct." It is a psychometrically calibrated benchmark that accounts for slight differences in difficulty between exam forms. Two candidates who sit different versions of the AIGP can be compared on equal terms because of this system.

The IAPP uses a process called psychometric scaling to produce this number. An exam development board establishes a "Cut Score" — the raw number of correct answers required — for each version of the exam. That raw number may shift slightly if one exam form is marginally harder than another. The 300 scaled score is the consistent output of that process. It is what you see on your score report; it is what determines whether you pass.

What You're Actually Being Scored On

The exam contains 100 multiple-choice questions, but not all of them count. The structure is as follows:

• 85 scored items — these determine your result.
• 15 unscored pretest items — statistical placeholders for future exam development. They look identical to scored questions; you will not know which is which.

Each correct answer is worth exactly one point. There is no partial credit, and — critically — there is no penalty for wrong answers. An unanswered question counts as incorrect. The strategic implication is absolute: never leave a question blank. A guess gives you a 25% chance of a correct answer. A blank guarantees zero.

With 85 scored items and a passing threshold that scales to 300, most candidates should target roughly 60–65 correct answers on the scored portion as a practical goal — though the exact raw figure varies slightly by exam form. Build your preparation for a comfortable margin above that floor, not for the minimum.

How the Exam Is Structured

You have 3 hours total, broken down as follows: 2 hours and 45 minutes (165 minutes) of active testing time, plus one optional 15-minute break mid-exam. When you take the break, the clock stops — but you cannot return to questions you have already passed. This is not a minor detail. It means your decision to break must be deliberate.

At 100 questions across 165 minutes, your baseline pace is approximately 1 minute and 36 seconds per question. That is tight. What makes it tighter: roughly 30% of the exam consists of complex scenario-based case studies, which consistently consume more time than standalone questions.

The recommended approach for case studies is to scan the scenario first for structural identifiers — specifically, whether the organization described is an AI developer or an AI deployer. This single determination resolves a significant portion of what Domains III and IV will ask you. Misidentifying the organizational role is the most common source of missed points on the practical portion of the exam.

Where the Points Are: Domain Weighting

The current exam is built on Body of Knowledge Version 2.1, effective February 2026, which reflects the full regulatory landscape including the EU AI Act and ISO/IEC 42001. The blueprint weights four domains as follows:

Domains III and IV together account for up to 50 of the 85 scored questions — nearly 60% of your result. These are also the domains most closely tied to practical application and scenario reasoning, not memorization. Candidates who over-index on Domains I and II (which are more definitional in nature) and under-prepare for the operational governance tasks in III and IV consistently fall short.

Exam Logistics and Pricing

The AIGP is administered through Pearson VUE, available at physical testing centers or via the OnVUE remote proctoring platform. Pricing varies by IAPP membership status:

One logistical detail worth highlighting: the discounted retake fee of $475 also applies if you already hold another active IAPP certification, such as a CIPP/E or CIPM. If you are an existing IAPP certificant sitting the AIGP for the first time, confirm your eligibility for this rate before registering.

How to Actually Prepare for a 300+

The AIGP is not a recall exam. The IAPP explicitly designs questions around Bloom's Taxonomy, and the upper domains — where the bulk of your points sit — operate at the Apply and Analyze levels. You should expect questions that ask you to differentiate, execute, implement, relate, and solve, not merely define or list.

IAPP official training materials provide a necessary foundation, but they are not sufficient on their own. The exam is built to assess competence, not course completion. Supplement your preparation with primary sources: the NIST AI Risk Management Framework and the text of the EU AI Act are non-negotiable reading for any candidate targeting a comfortable pass margin.

The single most high-leverage habit you can build is the ability to rapidly and correctly classify the organizational role in any scenario. Every time a case study is placed in front of you, your first move should be: is this entity a developer or a deployer? That classification governs which governance obligations apply, which domain concepts are relevant, and which answer choices are plausible. Master that reflex, and Domains III and IV become significantly more tractable.

How the Cut Score Is Actually Set

It's worth understanding, at a high level, why the raw number of correct answers needed can shift slightly between exam forms while the scaled passing score stays fixed at 300. Professional certification exams like the AIGP typically rely on a formal standard-setting process: a panel of subject matter experts reviews each question and judges how a "minimally qualified candidate" — someone who just barely has the competence the credential is meant to certify — would be expected to perform on it. That judgment process produces the raw cut score for a given exam form.

Because different exam forms draw from different question pools with slightly different average difficulty, the raw cut score isn't identical across forms — but the scaled score is deliberately engineered to be. This is standard practice across serious professional certifications, not something unique or unusual about the AIGP. It exists specifically so that "I passed with a 320" and "you passed with a 340" reflect genuinely different performance levels, regardless of which exam form either of you actually sat.

Misconceptions That Actually Cost Candidates Points

A few specific misunderstandings about the scoring system show up repeatedly among candidates who don't pass on their first attempt — and they're avoidable once named clearly:

• Treating 300/500 as "you only need 60%." As covered above, this arithmetic doesn't apply to a scaled score. Candidates who study to a literal 60% comprehension level across all material, rather than building genuine competence, are calibrating against the wrong number.
• Assuming every domain carries equal weight. Domains III and IV together outweigh Domains I and II by a meaningful margin. Spending equal study time across all four domains under-invests in the two domains that determine most of your outcome.
• Skipping unfamiliar-looking questions to "come back later." Given the tight per-question pacing and the irreversible nature of the optional break, questions skipped and not revisited effectively become blanks — which, as covered above, guarantee zero points rather than the 25% chance a guess provides.
• Believing pretest items don't matter enough to think carefully about. Since you cannot identify which 15 of the 100 questions are unscored, treating any question with less rigor than another is a gamble against your own scored result.

Frequently Asked Questions