Defensible Cost Contingency
Reviewed September 2026 · Civil Systems LLCA contingency number is only as defensible as the risks and assumptions behind it.
A flat percentage can be useful as an early allowance, but it does not explain which uncertainties drive the amount. This tool takes a risk-based approach: it models the cost impact of entered events and asks how often the simulated total stays within a given budget.
Some owners set a confidence threshold through policy. For example, Federal Transit Administration reviews of major capital projects use P65 cost and schedule estimates. That does not make P65 a universal answer. Another owner may choose a different percentile based on the decision being made, its tolerance for overrun, and the quality of the model.
Use the interactive tool below to build a deterministic baseline and risk register. For each event, enter the conditional Low, Most Likely, and High cost impact alongside its probability. Adjust the skew to prefill the Most Likely value, then run the simulation to calculate statistical targets.
| Risk / Uncertainty Item | Distribution Skew | Low Estimate ($) | Most Likely ($) | High Estimate ($) | Prob (%) |
|---|
How to Read the Results
Frequency histogram (blue bars)
This shows the spread of cost impacts from the risk events you entered across 10,000 simulated runs. Taller bars mark ranges that appeared more often in this model. A wide spread means the entered risks produce a wider range of costs; it does not prove that every source of project uncertainty has been captured.
S-curve (green line)
This represents cumulative probability: the percentage of simulated runs at or below a given amount. If the S-curve crosses 80% at $150k, then 80% of the outcomes produced by this model were $150k or less. Risks omitted from the inputs are not covered by that statement.
Why Practice Beats Formulas
Running this simulator once teaches you the mechanics. Running it against a live, moving project, where the risk you deprioritized last week comes back this week, is what Critical Path does differently. The percentile is the same idea. The stakes just don't reset every time you reload the page.
How much contingency should a project include?
There is no universal percentile. The target should follow the owner's policy, decision context, risk tolerance, and the quality of the underlying model. FTA major capital project reviews use P65 cost and schedule estimates, but that is a program-specific requirement rather than a rule for every project.
What's the difference between P50, P65, and P80 contingency?
Each percentile is a result from the model. At P80, 80% of simulated outcomes are at or below that amount and 20% are above it. The statement is only as reliable as the risks, impacts, dependencies, and assumptions represented in the model.
Why not just use a flat percentage for contingency?
A flat percentage may be suitable for a rough early allowance, but it does not show which uncertainties drive the amount. A risk-based model connects contingency to stated events, likelihoods, and conditional cost impacts while making its assumptions visible.
A percentile is one decision point.
Critical Path continues from there: decide which risks to address, spend limited resources, and work with the consequences later.