PERT — Program Evaluation and Review Technique
PERT is the three-point-estimate counterpart to the Critical Path Method. Where CPM assumes one duration per activity, PERT takes three — optimistic, most likely, pessimistic — and calculates a weighted expected duration plus a variance. Developed by the US Navy in 1958 for the Polaris programme, PERT is still useful on modern construction projects, but almost always as a sizing technique inside a CPM master, not as a replacement for it.
What is PERT?
PERT (Program Evaluation and Review Technique) is a network-based scheduling method that computes an expected activity duration from three estimates — optimistic (O), most likely (M) and pessimistic (P) — using the weighted-mean formula Expected = (O + 4M + P) / 6, and a variance from ((P − O) / 6)². Expected durations feed into a network diagram the same way CPM does; variances along the critical path give a probabilistic view of the overall project duration.
Also known as: three-point estimation, weighted-mean estimation, program evaluation and review technique. The acronym PERT is universal; the full name is used mostly in academic references.
The PERT formulas
Two formulas carry the method. The first gives an expected duration; the second quantifies how much uncertainty is around it.
Expected duration
TE = (O + 4M + P) / 6
A weighted mean. The most-likely estimate carries four times the weight of either extreme, which reflects that modal duration is where the probability mass sits.
Activity variance
σ² = ((P − O) / 6)²
The spread between optimistic and pessimistic divided by six, squared. Variances along the critical path sum to give project variance; the square root gives the standard deviation.
- Optimistic (O): the best credible duration — roughly a 1-in-100 chance of beating it. Not fantasy; not aspiration; a defensible best case.
- Most likely (M): the modal duration, the one you would plan for in a normal CPM master.
- Pessimistic (P): the worst credible duration — roughly a 1-in-100 chance of being that bad. Not catastrophe; a defensible worst case.
Worked example: sizing a soft-ground tunnelling drive
A tunnelling contractor is estimating a 300 m drive through variable ground. Historical data for the crew and the TBM gives useful bounds, but the ground survey carries real uncertainty. Three estimates in working days: O = 24, M = 36, P = 72.
| Value | Working | Result |
|---|---|---|
| Expected (TE) | (24 + 4·36 + 72) / 6 = (24 + 144 + 72) / 6 | 40 days |
| Variance (σ²) | ((72 − 24) / 6)² = 8² | 64 |
| Standard deviation (σ) | √64 | 8 days |
| TE ± 1σ (≈68% range) | 40 ± 8 | 32 – 48 days |
| TE ± 2σ (≈95% range) | 40 ± 16 | 24 – 56 days |
What the number means
The expected duration of 40 days is longer than the most-likely estimate of 36 — the pessimistic tail is long enough to shift the weighted mean. Standard deviation of 8 days means the activity realistically finishes somewhere between 32 and 48 days about two-thirds of the time, and between 24 and 56 days about 95% of the time. That is a very different planning instruction than “36 days” alone.
If this activity is on the critical path, its 64 variance contributes to the project-level variance and the planner can size a defensible contingency. If it is not on the critical path, the planner can check how much float protects downstream work against a pessimistic outturn.
PERT vs CPM in modern construction planning
PERT and CPM were developed independently within a year of each other (PERT 1958, CPM 1957) and are often treated as alternatives. In modern construction they are better treated as complementary, with each one appropriate to a different kind of activity.
| Dimension | PERT | CPM |
|---|---|---|
| Estimate per activity | Three-point (O, M, P) | Single deterministic duration |
| Output | Expected duration + variance | Longest-path duration |
| Best for | First-of-a-kind, uncertain or novel activities | Known, repetitive, mature-method activities |
| Treats uncertainty as | Explicit, quantified | Implicit, absorbed in float |
| Modern use in construction | Sizing uncertain activities inside a CPM master | Master schedule for the whole programme |
| Combines well with | Lean production control (LPS) for weekly reliability | Lookahead and pull planning for operational control |
Where PERT breaks down on a real construction site
PERT is mathematically clean. The four failure modes below are the reasons it is rarely used as a full master schedule method on modern construction projects, and why most teams treat it as a sizing technique rather than a scheduling system.
The three estimates are still judgement calls
O, M and P are themselves estimates. If the team has no genuine basis for them — no historical data, no comparable work — PERT just gives a more respectable-looking number for the same guess. Garbage in, formula out, garbage out.
The independence assumption is often false
The variance maths assumes activity durations are statistically independent. On construction sites they rarely are — one trade slipping cascades into the next trade's labour availability, which cascades into materials storage, which cascades again. Real project variance tends to be worse than the PERT sum predicts.
A single expected duration implies more precision than the method provides
Reporting “expected 40 days” invites the project to treat 40 as a target rather than a central estimate. The variance that makes PERT useful is often dropped from the summary that reaches leadership, and the method collapses back to a single-point CPM in all but name.
No feedback loop from site reality
PERT estimates are made once at planning. If the pessimistic case materialises in week two, PERT does not update itself. Modern lean construction pairs any front-end estimation method with the Last Planner System, which measures commitment reliability every week and feeds that back into the next lookahead.
How VisiLean handles PERT-style uncertainty
VisiLean Planner runs a CPM master because that is what every integration (P6, MS Project, Asta) and every modern construction tool is built around. PERT-style uncertainty at the front-end estimate is best managed through what happens after the estimate: the Last Planner lookahead and the weekly PPC loop keep the plan honest as reality arrives, so the estimate does not have to have been correct in the first place.
- CPM master imported directly from Primavera P6, Microsoft Project or Asta Powerproject — no re-authoring to a different scheduling philosophy.
- The six-week lookahead screens upcoming activities for the constraints that typically drive variance in the first place — design, materials, prerequisites, access, inspections.
- Weekly PPC and reason codes build a feedback loop on the front-end estimates: if a trade consistently finishes faster or slower than planned, the trend is visible before the next lookahead.
- Progress updates captured from the field via LiveSite give a continuous planned-vs-actual view, so variance is tracked as a trend rather than reported once as a static number.
Related methods
Critical Path Method (CPM)
The sister technique developed in parallel with PERT in the late 1950s. CPM is the master-schedule method; PERT sits inside it to size uncertain activities.
Read moreLast Planner System
Modern production control. PERT estimates uncertainty front-end; LPS manages it week by week through lookahead screening and PPC feedback.
Read moreLookahead Planning
The six-week screening step that catches constraints before they turn an optimistic PERT estimate into a pessimistic actual.
Read moreVisiLean Planner
The product that runs a CPM master, screens the lookahead and tracks PPC — turning front-end estimates into a live, correcting schedule.
Read moreFrequently asked questions
What is PERT in project management?
PERT stands for Program Evaluation and Review Technique. It is a scheduling method that calculates an expected activity duration from three estimates — optimistic, most likely and pessimistic — using a weighted-mean formula: Expected = (O + 4M + P) / 6. PERT was developed by the US Navy in 1958 for the Polaris submarine programme to deal with activities whose true duration was uncertain. It is one of the two foundational network-planning techniques, alongside the Critical Path Method.
What is the PERT formula?
Expected duration (TE) = (O + 4M + P) / 6, where O is the optimistic estimate (best case, roughly 1-in-100 chance), M is the most likely estimate (modal duration), and P is the pessimistic estimate (worst case, roughly 1-in-100 chance). The variance of each activity is ((P − O) / 6)². Variances along the critical path sum to give the variance of the overall project duration — the main reason PERT exists at all.
What is the difference between PERT and CPM?
CPM uses a single deterministic duration per activity and calculates the longest path. PERT uses three estimates per activity and calculates an expected duration plus a variance. CPM is appropriate when activities have well-understood durations — repetitive construction, defined scope, mature trades. PERT is appropriate when durations are genuinely uncertain — first-of-a-kind work, new sites, novel methods. Most modern construction planning uses CPM for the master schedule and treats PERT as a technique for sizing specific uncertain activities inside it, not as a replacement for CPM.
When should construction teams use PERT instead of CPM?
Rarely as a full replacement for CPM, and more often as a sizing technique for individual activities whose duration is genuinely uncertain — a soft-ground tunnelling drive, an unprecedented commissioning sequence, early-works ground investigation, a first-of-a-kind façade installation. The three-point estimate gives a defensible duration and variance where a single deterministic value would be a guess. For most routine construction activities with established crews and known methods, CPM alone is sufficient and the PERT overhead is not justified.
What are the limitations of PERT in construction?
Three big ones. First, the three estimates are themselves judgement calls — garbage in, garbage out. Second, the formula assumes activity durations are independent, which is often false on a construction site where one trade's slip cascades. Third, PERT reports a single expected duration that implies more precision than the method actually provides. Modern practice is to treat PERT as a sizing technique within CPM, and to combine it with a reliable production-control layer like the Last Planner System rather than relying on PERT alone to manage uncertainty.
See a CPM master run with lean production control
VisiLean imports your Primavera, MS Project or Asta master and runs it with the Last Planner lookahead and PPC on top — the front-end estimate becomes a live, correcting plan.