CPA AUD Exam Format

AUD at a glance

AUD is a four-hour computer-based section of the Uniform CPA Examination. In the 2026 AICPA blueprint, you answer 78 multiple-choice questions (MCQs) and 7 task-based simulations (TBSs). Five testlets organize the work. You complete the testlets in order, submitting each before the next one opens.

The count describes separate items, not equal units of work. A simulation may ask you to complete several response fields using one or more exhibits. Another may have fewer fields but require close reading or calculations. Do not budget time by multiplying seven simulations by a fixed number of minutes without seeing their demands.

What happens in each testlet

The two MCQ testlets

The first two testlets each contain 39 questions. Questions present a short fact pattern or direct prompt with answer options. They can test terminology, standards, ethics, engagement selection, assertions, risk, evidence, audit procedures, reporting, or a calculation. Some questions are straightforward recall. Others require you to distinguish between two plausible procedures or identify the conclusion that follows from the facts.

You can move among questions within the open testlet and review answers before submitting it. Once submitted, you cannot return to that testlet. That makes a simple working method useful: answer the items you can solve, mark uncertain ones if the software permits, then revisit them before you submit. Avoid sacrificing several answerable questions to one stubborn item.

The exam is not an invitation to guess what a testlet says about your performance. A second set that feels difficult does not reliably tell you how you are doing. Item design and scoring are controlled by the exam, and your impression is an unreliable substitute for answering the question in front of you.

The three simulation testlets

The final three testlets contain seven TBSs in total. A task-based simulation is a case-based problem that can include records, correspondence, reports, spreadsheets, or other exhibits. You may need to select documents, classify findings, complete a reconciliation, calculate an amount, identify an assertion, or choose the next audit procedure. The mix depends on the task.

A TBS tests whether you can use information, not just recognize a phrase. Read the task first so you know what decision is required. Then inspect the exhibits that bear on that decision. Record the relevant facts, ignore material that does not affect the task, and apply the applicable audit concept. If a cell asks for an amount, show your calculation on scratch work and check signs, units, and rounding before entering it.

Not every exhibit is necessarily relevant to every response. Treat each exhibit as evidence whose source and purpose matter. A client-prepared schedule may help locate a population, but the auditor still considers its completeness and accuracy. A third-party document may corroborate a detail, but it does not automatically prove every related assertion. Tie the document to the precise response requested.

Timing and breaks

Four hours is the examination testing time. NASBA’s current FAQ says you are offered a standardized 15-minute break about midway through the section, after the first TBS testlet. You may accept or decline it. If accepted, it does not count against testing time. Other breaks may be available between testlets, but the clock continues during them. The appointment itself includes time beyond the four hours for check-in and post-exam steps, so the appointment duration is not extra answer time.

A planning average can help you notice when you are stuck, but it cannot tell you how long each item deserves. Dividing four hours by 85 listed items yields roughly 2.8 minutes per item, but that average is misleading if applied to every item: a short MCQ and a multi-exhibit simulation are not comparable units. Think instead in stages. Keep moving on MCQs, reserve meaningful time for simulations, and monitor the clock after each testlet.

Use optional breaks deliberately. The standardized break is a natural chance to reset before the remaining TBS work. For other pauses, weigh the need to rest against the time they consume. Follow the test center’s instructions and the current Candidate Guide; access to personal materials or devices during breaks is restricted.

How the score is weighted

AICPA reports the section result on a scale from 0 to 99, with 75 as the passing score. A score of 75 does not mean 75 percent correct. For AUD, MCQs contribute 50 percent and TBSs contribute 50 percent. This makes it a mistake to study only MCQs because they look familiar or to treat simulations as an optional final-week activity.

AICPA explains that component scores are scaled using formulas that take account of correctness and relative item difficulty. The score is not curved. A raw practice percentage from a third-party question set cannot be converted directly into an official AUD score. Use practice to identify knowledge and reasoning gaps, not to promise a particular scaled result.

Some questions are pretest items and do not count toward a candidate’s score. They are mixed with operational items and are not identified to candidates, so there is no useful way to spot and skip them. Work every question as if it counts.

Original format example

Consider this original multiple-choice style scenario: The auditor selects a sample of recorded year-end sales and traces each entry to a customer order and carrier evidence. Which assertion is most directly addressed by starting with the recorded sales entries? The best answer is occurrence. The direction is from recorded transactions to supporting evidence, asking whether the recorded sales happened. Starting with shipping records and tracing into the ledger would instead focus more directly on completeness.

A simulation might expand that same issue. You could receive a sales listing, several shipping records, contract terms, a credit memo, and a year-end journal entry, then be asked to identify cutoff exceptions and calculate the adjustment. The important skill is not memorizing that every shipping record controls revenue recognition. Read the contract and facts to determine when control transfers, then connect that timing to the entry.

What to practise before test day

Use the AICPA sample test to learn the look and feel of the software and available tools. AICPA describes it as a shorter, unscored practice experience with questions from all six exam sections. It is not a readiness assessment and it may not match every live software or security detail exactly. Treat it as interface orientation, not content coverage.

For content practice, combine MCQs with original simulations based on the blueprint. After an MCQ, explain why the correct answer fits and what fact would make the nearest distractor correct. After a simulation, review whether you misread an exhibit, selected the wrong assertion, made an arithmetic error, or applied the wrong reporting rule. Those are different problems and need different practice.

Practise the mechanics of a simulation: read the requested output, scan the exhibit titles, identify relevant records, and reconcile the evidence before you choose or enter a response. Work with tables and spreadsheet-like information rather than studying only prose. Check formulas independently when possible. If two exhibits conflict, note the conflict instead of averaging the information or choosing the more polished-looking document.

Timed practice should include both question types and the planned break. Do not turn every practice set into a race. First become accurate; then reduce avoidable delays such as rereading without a question, hunting for a rule you already know, or repeatedly recalculating the same amount. A useful target is a consistent method under time pressure, not a dramatic pace on one unusually easy set.

Common format questions

How many MCQs are on AUD? The 2026 blueprint lists 78: 39 in each of the first two testlets.

How many AUD simulations are there? There are 7, arranged across three testlets containing 2, 3, and 2 simulations.

Can I return to earlier testlets? No. You can review items within the active testlet before submitting it, but you cannot go back after submission.

Does the 15-minute break use my exam time? NASBA says the standardized break after the first simulation testlet does not count against testing time. Other optional breaks do count.

How to use the four hours

There is no single perfect minute-by-minute schedule for every candidate. The testlets differ in size and task, and candidates differ in reading speed and strengths. Build a plan around checkpoints instead of trying to assign an identical number of minutes to every question. For instance, decide in advance when you will check the clock during the two MCQ testlets and after the first TBS testlet. If you are behind, shorten review of questions where you have already chosen a defensible answer rather than rushing every remaining simulation.

The arithmetic average of four hours over 85 listed items is about 2.8 minutes per item. Do not mistake that result for a quota. MCQ count and TBS count are not equivalent: one simulation can contain several linked response fields and several exhibits. A better pacing question is whether the time spent is proportionate to the reasoning required. If an MCQ is consuming several minutes without progress, mark it if the interface allows and return before submitting the testlet.

At each checkpoint, ask three things: have I answered the current testlet, have I left time to review uncertain items, and am I protecting enough time for the later simulations? A checkpoint is not a reason to panic if the clock differs from a practice run. It is a prompt to make a realistic adjustment. Do not borrow time from every later testlet for one early question.

What the software tools change

The examination is computer delivered, so practice should include reading on screen and entering responses in the available formats. A spreadsheet function may help you sort, filter, or calculate from data when the task calls for it. A calculator may help with arithmetic. Neither tool identifies the relevant assertion or tells you whether a source is reliable. Decide what the task is asking before opening tools or manipulating a schedule.

AICPA provides a sample test and tutorial material so candidates can become familiar with the testing software. The sample experience uses questions from all six sections and is shorter than the actual exam. It is not scored and does not determine readiness. Software familiarity matters because searching for a function during a case can consume attention, but practicing the interface cannot replace learning auditing concepts.

Use the sample environment to try the types of navigation and response controls it actually demonstrates. Do not assume that every display setting, keyboard behavior, or security restriction will be identical in a browser sample and a Prometric test center. AICPA notes that the sample test may not operate exactly like the test-center CPA Exam Driver. Follow the current candidate instructions for your appointment.

Separate exam counts from scored weight

The count of 78 MCQs and 7 TBSs is not the scoring formula. AICPA assigns half the AUD score to the MCQ portion and half to the TBS portion. Consequently, an MCQ is not automatically worth the same share as one simulation. Simulations can have multiple response parts, and the score is scaled rather than calculated as a visible raw percentage. Focus on completing accurate work in both portions instead of trying to reverse-engineer an exact point value for a question.

Some items are pretest questions that are not used in scoring. They are mixed into the exam and are not marked. This has two practical consequences: do not spend energy trying to identify them, and do not treat an odd-looking prompt as optional. Respond with the same care you would give any other question.

Decide what to do when an item stalls

When a question becomes unproductive, pause long enough to name the missing piece. Is the problem that you cannot identify the assertion, do not remember a rule, cannot reconcile two records, or have misread the task? If you can isolate the missing piece, make the best supported response and move on. If the software permits marking for review, use that feature rather than rereading the same paragraph without a new question in mind.

For a simulation, do not start by reading every exhibit word for word when the task has not yet been identified. Read the requirements first, note the requested output, then scan the exhibit list for relevant information. Return to documents as needed and retain their context. An invoice, confirmation, and management explanation may relate to one account but support different assertions and carry different evidential weight.

A useful final check is narrow: Did you respond to every required field? Did you reverse an amount by mistake? Did you use the requested unit or date? Does the response answer the task rather than a nearby issue? This catches avoidable omissions without inviting you to reopen every judgment you already made.