How Many Practice Questions Do You Need Before the PMP Exam?
Ask this in any PMP study group and you will get a number. Usually somewhere between 1,000 and 2,000 practice questions, plus three to five full-length mock exams, with a rule of thumb that you should be scoring above 70 to 75 percent on fresh mocks before you book.
Those numbers are not wrong, exactly. They are a reasonable description of what people who passed happened to do. But they get repeated as a target, and treating them as a target is how candidates end up grinding through 2,500 questions and still failing, while somebody else passes on 900.
Where the Common Numbers Come From
The 1,000 to 2,000 range is essentially a survey result. It is what passing candidates report, aggregated across forums and course completion data. That makes it a useful sanity check and a terrible plan.
The problem is survivorship plus confounding. People who did 1,500 questions also typically spent eight to twelve weeks studying, read the process material at least once, and reviewed their mistakes. The question count is the most easily measured part of that bundle, so it became the headline number. It is a proxy for effort, not a cause of passing.
The mock exam guidance is on firmer ground. Full-length timed mocks do something the question count does not: they test whether you can maintain judgment for 230 minutes across 180 questions. Stamina and pacing fail a meaningful number of otherwise-prepared candidates. Three to five full mocks is genuinely good advice, and it is advice about a different thing than question volume.
The Number That Actually Predicts Your Score
Not how many questions you have answered. How many distinct reasons you have for getting questions wrong.
Here is the mechanism. The PMP exam is mostly situational: it describes a scenario and asks what you do next, and the answer options usually contain two that a reasonable project manager could defend. You do not get those wrong at random. You get them wrong because you hold a small number of specific wrong instincts — you escalate when you should consult, you act when you should gather information, you reach for a tool when the situation calls for a conversation.
Most candidates have between four and eight of these. They generate almost all the errors. And they do not go away from volume, because doing another 300 questions with the same instinct just produces the same error 40 more times.
The Review Practice That Finds Them
For every question you get wrong, write one sentence. Not a summary of the correct answer — the answer key already told you that. Write why the answer you picked looked right.
Examples of what that sentence looks like when it is doing its job:
- "I picked 'inform the sponsor' because the issue felt serious, but nothing had actually blocked the work yet."
- "I picked 'update the risk register' because it was the most process-correct option, and the question was asking what to do first, not what to do eventually."
- "I picked the option where the PM decides, because the team's proposal looked wrong to me."
After thirty or forty of these, read them all in one sitting. The clustering is usually obvious and slightly embarrassing. You will find that you escalate too early, or that you consistently choose the documentation option when the exam wants a conversation, or that you treat "first" and "best" as the same word.
That list is your real study plan. It is worth more than the next thousand questions.
So How Many Questions, Concretely
If you want a number to plan around, here is a defensible one, with the reasoning attached so you can adjust it.
| Phase | Questions | Purpose |
|---|---|---|
| Diagnostic | 50-100 | Find your starting level before you study anything |
| Topic drilling | 600-1,000 | Build the reflexes, in small filtered sets by domain |
| Mock exams | 540-900 | 3 to 5 full-length timed, for pacing and stamina |
| Targeted re-drill | 200-400 | Only your identified failure patterns |
That totals roughly 1,400 to 2,400, which lands right on the conventional range — but the composition is the point, not the total. Most of the difference between candidates who pass comfortably and candidates who do not is whether that third and fourth row happened at all.
A note on the diagnostic: take it before you study, not after. A cold score of 45 percent tells you which domains to weight. A score taken after four weeks of reading tells you much less, because by then you have no baseline to compare against.
Two Sample Schedules
Four weeks, working full time
| Week | Focus | Volume |
|---|---|---|
| 1 | Diagnostic, then drill your two weakest domains | 50 diagnostic + 250 drilled |
| 2 | Remaining domains, agile and hybrid emphasis | 350 drilled |
| 3 | Two full-length timed mocks, deep review of each | 360 + review |
| 4 | Re-drill only your failure patterns, one final mock | 200 + 180 |
Eight weeks, working full time
| Week | Focus | Volume |
|---|---|---|
| 1 | Read the Examination Content Outline, diagnostic set | 100 |
| 2-3 | People domain and agile ceremonies, small daily sets | 400 |
| 4-5 | Process domain, procurement, risk, change control | 500 |
| 6 | Business environment, first full mock | 100 + 180 |
| 7 | Two more full mocks, written review of every miss | 360 + review |
| 8 | Failure-pattern re-drill, taper, final mock | 300 + 180 |
Both schedules put the first full-length mock later than most plans do. That is deliberate. A full mock taken before you have the basic reflexes mostly measures how much you do not know yet, which you already knew, and it burns one of the few genuinely fresh 180-question sets you have available.
When You Are Stuck, Do Not Add Volume
The most common failure mode: practice scores plateau around 65 percent, so the candidate buys a second question bank and does another 800 questions.
If your score has been flat across two mocks, more questions is almost certainly the wrong move. A plateau means your errors are systematic, and systematic errors are immune to volume. Go back to your wrong-answer sentences. If you have not been writing them, write them for the next 40 misses before you do anything else.
The other thing worth checking when scores stall: are you reading the question stem for what it actually asks? "What should the project manager do first" and "what is the best response" and "what should the project manager have done" are three different questions, and a surprising share of misses are candidates answering a slightly different question than the one on the screen.
Practical Guidance on Pacing
The real exam gives you 230 minutes for 180 questions, which is about 76 seconds each, with two scheduled breaks. Most people finish with time left, but the people who run out tend to do so because they spent four minutes on a handful of ambiguous items early on.
Practice with a clock from the start. If a question has taken you more than two minutes, mark it, pick the option that best matches your default reflexes, and move. You will usually be right, and if you are not, one question costs you far less than the twelve minutes you were about to spend.
Also, do at least one full mock at the actual time of day your exam is scheduled. Cognitive performance is not constant across the day and an 8am exam is a different event than a 6pm one.
The Short Answer
Enough questions to have seen every question type at least a few times — call it 1,200 to 1,500 — plus three to five full-length timed mocks, plus a written record of why you got things wrong. The written record is the part that most candidates skip and the part that does the most work.
If you are still at the beginning of this, get a baseline first. Take 50 free practice questions with no card required, sort your misses by domain, and let that decide where the next month goes.