The PERT formula has exactly two inputs that trip candidates: the 4 and the 6. Most people get the three-point structure right. They add optimistic, most likely, and pessimistic. Where they lose points is the multiplication and the divisor, and not knowing when the exam wants the triangular average instead.
What PERT actually is
PERT stands for Program Evaluation and Review Technique. On the exam, it means one thing: a weighted three-point estimate that gives extra weight to the most likely scenario.
The formula: (O + 4M + P) / 6
O = Optimistic estimate M = Most likely estimate P = Pessimistic estimate
The 4 reflects a core assumption: the most likely outcome is four times more probable than either extreme. The 6 in the denominator is the sum of those weights (1 + 4 + 1). You divide by 6 to get a single estimate.
Example: A task has an optimistic estimate of 4 days, most likely of 7 days, and pessimistic of 16 days.
PERT estimate = (4 + 4×7 + 16) / 6 = (4 + 28 + 16) / 6 = 48 / 6 = 8 days
Without the 4, you get (4 + 7 + 16) / 3 = 9 days. That is the triangular average, which the exam calls a “simple average” or “triangular distribution.”
Two different answers, two different formulas. The question tells you which one to use.
When to use PERT vs the simple average
Both distributions appear on the PMP. The signal is in the question stem.
Use PERT (beta distribution) when: - The question says “three-point estimate” without naming a distribution - The question asks for the “expected duration” and gives three estimates - The question references a “weighted average of three-point estimates”
Use triangular (simple average) when: - The question explicitly says “triangular distribution” - The question asks to “average the three estimates equally”
If the question gives you three time estimates and asks for a duration with no distribution named, default to PERT. PMI treats PERT as the standard three-point method on the large majority of questions.
The standard deviation and variance formulas
PERT does not stop at a single estimate. It comes with two more formulas that appear on the exam.
Standard Deviation (SD) = (P - O) / 6
This measures the spread of uncertainty around the PERT estimate. A wider spread means less confidence.
Using the same example: SD = (16 - 4) / 6 = 12 / 6 = 2 days
Variance = SD squared = [(P - O) / 6]^2
Variance = 2^2 = 4 days squared
When do these show up? Schedule range questions. PMI may give you three tasks on a critical path and ask for the total standard deviation of the path. You add the variances (not the standard deviations), then take the square root.
Path total variance = variance of task 1 + variance of task 2 + variance of task 3 Path standard deviation = square root of total variance
Candidates who add standard deviations directly get the wrong answer. This shows up on practice exams as a “close but wrong” choice.
Three traps the PMP sets on PERT questions
Trap 1: Forgetting the 4
The most common arithmetic error. A candidate in a rush writes (O + M + P) / 6 instead of (O + 4M + P) / 6. The 4M disappears and the answer is 1-2 days off.
Before calculating, write “4M =” on your scratch paper. Force yourself to multiply before you add.
Trap 2: Reporting SD when the question wants variance
A question asks: “What is the variance of this task’s duration estimate?”
Optimistic: 3 weeks. Most likely: 8 weeks. Pessimistic: 19 weeks.
SD = (19 - 3) / 6 = 16 / 6 = 2.67 weeks Variance = (2.67)^2 = 7.11 weeks squared
If the answer choices include both 2.67 and 7.11, you need to know which the question asks for. Variance always gets squared. SD does not.
Trap 3: Adding standard deviations across tasks
The exam gives you a two or three-task critical path and asks for the total range of completion time.
Wrong: add the standard deviations directly. 2 + 1.5 + 3 = 6.5 days.
Right: add the variances, then take the square root. (4 + 2.25 + 9 = 15.25, then sqrt(15.25) = 3.9 days.)
You do not need to know the statistics theory behind this. You need to remember: add variances, not standard deviations, then root the total.
A worked example end-to-end
Your project has two tasks on the critical path.
Task A: Optimistic 2 days, Most Likely 5 days, Pessimistic 14 days. Task B: Optimistic 1 day, Most Likely 4 days, Pessimistic 7 days.
Step 1: PERT estimates
Task A = (2 + 4×5 + 14) / 6 = (2 + 20 + 14) / 6 = 36 / 6 = 6 days Task B = (1 + 4×4 + 7) / 6 = (1 + 16 + 7) / 6 = 24 / 6 = 4 days
Step 2: Critical path duration
6 + 4 = 10 days total
Step 3: Standard deviations
Task A SD = (14 - 2) / 6 = 12 / 6 = 2 days Task B SD = (7 - 1) / 6 = 6 / 6 = 1 day
Step 4: Variances
Task A variance = 4 days squared Task B variance = 1 day squared
Step 5: Path variance and path standard deviation
Total path variance = 4 + 1 = 5 days squared Path standard deviation = sqrt(5) = 2.24 days
Step 6: The range question
If the question asks “In what range is there roughly a 68% chance the path finishes?” the answer is 10 ± 2.24 days, or 7.76 to 12.24 days.
(68% probability corresponds to ±1 standard deviation. 95% is ±2 SD. 99.7% is ±3 SD. The exam uses these probabilities directly.)
The quick decision card
When you see three duration estimates, ask two questions before touching any math.
Question 1: Does the question name a distribution? If it says “triangular,” use (O + M + P) / 3. If it says “beta” or “PERT,” or names nothing, use (O + 4M + P) / 6.
Question 2: Does the question ask for an estimate, a standard deviation, or a variance? Estimate = use PERT directly. Standard deviation = (P - O) / 6. Variance = square the standard deviation.
If it then asks about a path, add variances, not standard deviations, then take the square root for the path standard deviation.
Why candidates miss these questions even after studying
PERT questions are not a large chunk of the PMP exam. You will see 3-5 estimation questions total. But they are nearly always worth full marks or zero, because each question usually has one unambiguous correct answer and four plausible wrong answers.
The candidates who miss these questions share a consistent pattern: they know the formula but not the decision rules. They can write (O + 4M + P) / 6 on demand, but under time pressure they add standard deviations across tasks or report variance instead of SD.
That gap between recall and application shows up as a cluster on the PassCoach bias diagnostic. If your mock scores are inconsistent on quantitative questions, there is a good chance PERT is part of the pattern.
Join the waitlist to get early access when the diagnostic launches.
The formula is not the hard part. Knowing which number to report and when not to use it at all is where the points shift.