Full disclosure before anything else: we sell a PMP practice question bank, so we have an interest in you practising more. This article is about exam mechanics, and every figure below comes from PMI's 2026 PMP Examination Content Outline, effective July 2026.
Most pacing advice stops at "about 80 seconds per question." That average is correct and almost useless on its own, because it says nothing about the decision you actually face: this one is a coin flip between two defensible answers, do I settle it now or come back to it?
The answer depends on a rule that has nothing to do with time per question.
A warning about stale numbers, including ours
If you see 230 minutes, or 175 scored questions and 5 pretest items, or breaks described as falling "after question 60 and after question 120" - those are pre-July-2026 figures. They come from the PMP Certification Handbook PDF, which is still downloadable from pmi.org but is labelled the "January 2021 Exam Update" edition.
We know this because an earlier version of this very article used those numbers. We had read the handbook, seen a 2023 copyright, and treated it as current. It is not. The 2026 Examination Content Outline is the document that describes the exam you will actually sit.
Check the edition line on anything you are relying on. Half the prep articles online have not.
What PMI's 2026 outline actually says
- 180 questions, of which 10 are pretest questions. "Pretest questions do not affect the score." So 170 are scored.
- "The allotted time to complete the center-based exam is 240 minutes."
- "For the PMP certification exam, there are two 10-minute breaks in the exam. The first break occurs after the case-study section, and the second break occurs approximately midway through the independent question portion of the exam."
- And the sentence that should change how you work: "Once you review your responses and start your break you will not be able to return to the questions from the previous section of the exam."
- Domain weights: People 33%, Process 41%, Business Environment 26%.
The lock, not the flag, is what matters
Whatever the interface calls the review control, your ability to revisit a question expires. It expires when you start a break, and it expires for good.
So "I'll come back to it" is never an open-ended promise. It is a commitment to revisit that question before the next break, in whatever time you have managed to bank.
And note where the first lock falls: after the case-study section. Those linked scenario questions are sealed before you ever see the independent portion. If you were planning to let early uncertainty ride and resolve it later with a clearer head, the exam does not allow it.
The arithmetic that sets your mark budget
240 minutes across 180 questions is exactly 80 seconds per question.
Here is the part most people skip: if you spend the full 80 seconds on every question, you arrive at each break with zero minutes of review time. Review time is not allocated to you. It is only what you bank by answering faster than the average.
| Your average per question | Time banked across 180 questions | Questions you could revisit (at ~90 sec each) |
|---|---|---|
| 80 sec | 0 min | 0 |
| 75 sec | 15 min | ~10 |
| 70 sec | 30 min | ~20 |
| 65 sec | 45 min | ~30 |
Two caveats that make this stricter than it looks. First, a revisit costs more than a first read, not less - you have to reload the scenario before you can re-decide, so 90 seconds is a fair planning figure. Second, that banked time is not one pool. The lock splits it across segments, and you can only spend what you banked before the break you are approaching.
PMI does not publish how many questions are in the case-study section. So you cannot pre-plan an equal per-section budget the way you could with three 60-question blocks. What you can do is bank time continuously and keep your marked list short enough to clear before each break.
The practical consequence: your realistic budget is single digits per segment, not twenty.
What is worth a mark, and what is not
A mark is only useful if a second look can plausibly change the answer. That rules out more cases than people expect.
| Situation | Mark it? | Why |
|---|---|---|
| Two options both defensible; the difference is a judgement call | Yes | A second read often breaks the tie, because you re-anchor on what the question actually asks rather than on the option you liked |
| You do not know the concept at all | No | A review pass does not create knowledge you did not bring. Pick your best option and spend the time elsewhere |
| You misread the stem and caught yourself | No | Fix it now while the scenario is loaded. Marking it means paying the reload cost twice |
| You did arithmetic under time pressure | Yes | Re-checking a calculation is fast and the error rate is genuinely high |
| A case-study question you answered before reading later questions in the same scenario | Yes, and clear it first | Later questions in a linked scenario often reveal what the earlier one was driving at - and the whole case section locks at the first break |
| Very long stem you skimmed | Yes, once | On review, re-read only the final sentence - the actual ask - and check your answer against it |
The pattern: mark questions where the blocker is a decision, not questions where the blocker is knowledge.
Attach the mark to a reason, not to a feeling
You cannot write notes to yourself during the exam, so a mark carries no information beyond "something bothered me." By review time you will not remember what.
Fix this by only marking in the categories above, and naming it silently as you flag - "two-option tie," "re-check the maths," "re-read the ask." Three trained categories are recoverable thirty minutes later. A vague unease is not.
Train the review pass, not just the question count
Most candidates practise in sets of 20. That trains recall and nothing else - no pacing pressure, no mark budget, no review-pass discipline, and no experience of the lock.
Before exam day, run full-length timed sessions and treat the breaks as real: stop adding answers, take one review pass, then accept the break and do not look back. After each one, write down three numbers:
- How many questions you marked
- How many answers you changed on review
- How many of those changes were right
That third number should set your personal rule. If you consistently change more right answers to wrong ones than the reverse, your rule for exam day is: do not change an answer unless you can name the specific error in your first reading.
We have no study measuring the change-right rate for PMP candidates specifically, and we are not going to invent one. Measure your own - it takes a handful of full-length sessions and it is the most decision-relevant number you will collect.
Two rules for the day
Spend your review pass before you accept the break. PMI's wording ties the two together, and the lock closes behind it. A question you meant to revisit "after the break" is a question you have abandoned.
Do not spend the break thinking about a question you can no longer reach. Use it to reset your attention for what is still in front of you.
Practice questions to try next
These are free, public sample questions from our bank - no account needed:
- An important national project is in the execution phase. A time-constraint
- Just as a team reaches a mutual consensus on a preferred design, a
- During a requirements review session with a customer, a team member gets
If you want to rehearse pacing rather than single questions, try 50 questions free - no account required - or browse the full set of free sample questions.
Source: PMI, PMP Examination Content Outline (2026 edition, effective July 2026), pmi.org. Accessed 4 October 2026.