A probabilistic project-scheduling method that represents activities and precedence as a network while treating activity durations as uncertain variables (typically estimated via multi‑point expert inputs), uses those duration estimates to compute expected activity durations and variance measures, and derives probabilistic project completion times and likelihoods for schedules and critical paths.

Principle

When activity durations are uncertain, representing durations as probability distributions and propagating their moments through the precedence network yields expected project completion times and distributions of possible completion dates; criticality becomes a probabilistic attribute rather than a single deterministic path.

Demonstration

Illustrative scenario → A software release schedule contains tasks with uncertain durations. Recognition: the team supplies optimistic, most‑likely and pessimistic estimates for each task. Action: the analyst computes expected durations and variances using the three‑point estimates, aggregates through the network, and derives an expected project completion and probability of meeting a contractual deadline. Consequence: management obtains a quantified likelihood to inform contingency planning, resource buffering, or scope decisions while remaining aware that output quality depends on estimate reliability and independence assumptions.

Misapplication

Using PERT expected durations as fixed scheduling inputs without accounting for variance and correlation: the error is accepting the mean as representative of realized timing when the underlying distribution is skewed or when task durations are correlated, producing misleading schedule risk assessments.

Consequence

Correct application integrates uncertainty into planning, enabling probability‑based deadline assessment and targeted risk mitigation; incorrect use can produce false confidence, misplaced contingencies, underestimated variance, and suboptimal prioritization of risk responses because outputs depend heavily on the validity of duration estimates and independence assumptions.

Reversal

PERT’s probabilistic predictions are unreliable when input estimates are poor, when durations are strongly correlated across tasks, or when execution includes resource constraints that alter task concurrency; in such cases, Monte Carlo simulation, empirical statistical models, or resource‑constrained stochastic scheduling may be required to produce realistic risk assessments.

Boundary

Clearly within: projects with appreciable duration uncertainty where multi‑point expert estimates are available and precedence relations dominate timing. Boundary case: projects with moderate uncertainty but strong resource constraints—PERT may provide partial insight but must be combined with resource analysis. Clearly outside: deterministic scheduling where durations are known with high confidence, or RCPSP problems dominated by resource limits rather than duration uncertainty.

Semantic Tension

PERT sits between deterministic CPM and full Monte Carlo simulation: it aims to incorporate uncertainty with analytic approximations (often using three‑point estimates), but it trades off model simplicity against the richer outcomes and flexibility of simulation‑based approaches; choice reflects a trade between analytic tractability and fidelity to complex, correlated uncertainties.

Synthesis

PERT formalizes uncertainty in schedule estimates and converts subjective duration judgments into probabilistic statements about completion; its usefulness depends on estimate quality and independence assumptions, so it is most effective when combined with empirical validation, sensitivity analysis, or simulation for complex dependency structures.