How long will your project really take?
Turn best, likely and worst-case estimates into an expected time and a range you can trust.
What you enter stays on your device while you use this tool. Opening a shared link sends the settings in its address to HowToolbox to prepare the page.
Tasks
Expected time
16.3 d
The total can vary by about 2.1 d. It is 90% likely to be done within 19 d.
Ranges and spread
- 68% range
- 14.3 – 18.4 d
- 95% range
- 12.2 – 20.4 d
- 99.7% range
- 10.2 – 22.5 d
- Standard deviation
- 2.1 d
- Done by, 90% likely
- 19 d
- Done by, 95% likely
- 19.7 d
Turn best, likely and worst-case estimates into an expected time and a range you can trust.
How to use
- For each task, enter three estimates: the optimistic time if all goes well, the most likely time, and the pessimistic time if things go wrong.
- Add a row for every task done one after another, and pick the time unit.
- Read the expected total, the ranges around it and the time you can finish by with 90% or 95% confidence.
How it works
A single estimate hides how uncertain it is. The three-point (PERT) estimate asks for a best case, a normal case and a worst case, and weights the normal case most. The result is an expected time that is more honest than a guess, plus a measure of how much it can vary.
For optimistic O, most likely M and pessimistic P, the expected time is E = (O + 4M + P) ÷ 6 and the standard deviation is σ = (P − O) ÷ 6. For O = 2, M = 4 and P = 8 days, E = 26 ÷ 6 = 4.33 days and σ = 1. For tasks in a row, the expected times add up, and the variances (σ²) add up, so the total σ is the square root of that sum. About 68% of outcomes fall within E ± σ and 95% within E ± 2σ. Source: Wikipedia, "Three-point estimation".
The ranges are an approximation that treats the tasks as independent and the total as a normal curve. If tasks run in parallel, only the longest path counts, so enter just that path. Estimates still depend on honest inputs: the formula cannot fix an optimistic worst case.
Worked example
What is the expected time for a task estimated at 2, 4 and 8 days?
- Weight the likely time four times: (2 + 4 × 4 + 8) ÷ 6.
- The spread is the range divided by six: (8 − 2) ÷ 6 = 1.
The expected time is 4.33 days, give or take 1.
Method
No outside data is used: the result is worked out directly from the formula E = (O + 4M + P) / 6, explained under How it works.
Frequently asked questions
Why divide by 6 and weight the likely time by 4?
PERT approximates a beta distribution. Weighting the most likely time four times and each extreme once gives a mean close to that distribution, and the range from best to worst case covers about six standard deviations.
What is the difference between PERT and a simple average?
A simple average of the three numbers gives the extremes the same weight as the usual case. PERT trusts the most likely estimate more, so one wild worst-case number moves the result less.
How do I read the 68%, 95% and 99.7% ranges?
If your estimates were right, about 68% of the time the real total falls inside the first range, 95% inside the second and 99.7% inside the third.
What if tasks can run at the same time?
Enter only the tasks on the longest chain, the critical path. Tasks done in parallel do not add their times together.
Can I use it with money instead of time?
The maths works for any quantity with a best, likely and worst case, such as cost. The unit label only says days, hours or weeks, so ignore it and read the numbers as your own unit.





