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, not about our product, and the numbers below come from PMI's own PMP Certification Handbook, checked on 4 October 2026.
Most pacing advice for the PMP exam stops at "about 76 seconds per question." That average is correct and almost useless on its own, because it says nothing about the decision you actually face forty questions in: 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.
What PMI documents, and what it does not
From the PMP Certification Handbook:
- The exam "is comprised of 180 questions" — 175 scored and 5 pretest questions that do not count.
- "The allotted time to complete the center-based examination is 230 minutes."
- "For the PMP exam, there are now two 10-minute breaks in the exam. The first will appear after you complete questions 1-60 and review all of your answers. The second break will appear after you have completed question 120."
- And the sentence that matters most: "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."
Now the honest part. The handbook does not document a flag or mark-for-review button. Plenty of prep sites describe one, and the testing interface may well offer it, but we cannot point you at a PMI statement that says so. Treat the button as an interface detail to confirm in your practice environment and during your check-in walkthrough.
What PMI does document is the lock. And the lock, not the button, is what should change how you work.
The lock turns "I'll come back to it" into a deadline
Whatever the interface calls it, your ability to revisit a question expires three times during the exam: at question 60, at question 120, and at the end. Within a section you have one review pass. After you accept the break, the questions behind you are gone — not harder to reach, gone.
So every time you decide to come back to a question, you are not deferring it indefinitely. You are scheduling it into a review window whose length you have to earn.
The arithmetic that sets your mark budget
230 minutes across 180 questions is 76.7 seconds per question, or roughly 1:17. Spread evenly across three 60-question sections, that is about 77 minutes per section.
Here is the part most people skip: if you spend the full 76 seconds on every question, you arrive at the end of the section 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 | Minutes banked per 60-question section | Questions you can actually revisit (at ~90 sec each) |
|---|---|---|
| 76 sec | 0 min | 0 |
| 70 sec | ~6.7 min | ~4 |
| 65 sec | ~11.7 min | ~7 |
| 60 sec | ~16.7 min | ~11 |
A revisit costs more than a first read, not less — you have to reload the scenario before you can re-decide. Ninety seconds is a reasonable planning figure.
The practical consequence: your realistic budget is single digits per section, not twenty. If you mark 22 questions in section one and bank eleven minutes, you will not revisit 22 questions. You will revisit seven and feel the other fifteen as noise on your way to the break.
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 |
| 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. When you flag a question, name it silently — "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 section, not 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 at least four to six timed 60-question sections: 77 minutes, no pausing, make real mark decisions as you go, then take exactly one review pass and stop. 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 is the one that 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 four practice sections and it is the most decision-relevant number you will collect.
A 230-minute plan you can rehearse
| Section | Questions | Primary pass | Review pass | Then |
|---|---|---|---|---|
| 1 | 1-60 | finish by ~65 min | ~8 min | 10-min break |
| 2 | 61-120 | ~65 min | ~8 min | 10-min break |
| 3 | 121-180 | ~65 min | remaining time | end |
Three primary passes at 65 minutes plus three review passes at 8 minutes is 219 minutes, leaving roughly 11 minutes of slack across the whole exam. That slack is your margin for two or three genuinely hard questions, not a reserve you can spend early.
Two rules go with the table. First, spend your review pass before you accept the break, because the handbook's wording ties the break to having reviewed your answers, and the lock closes behind it. Second, do not spend the break thinking about a question you can no longer reach. Use it to reset your attention for the next sixty.
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 the 60-question rhythm rather than single questions, try 50 questions free — no account required — or browse the full set of free sample questions.
Source: PMI, Project Management Professional (PMP) Certification Handbook (pmi.org), accessed 4 October 2026.