A cost-benefit evaluation requires precise data on program outcomes. However, such data are unavailable when the analysis is prospective, and expensive and time-consuming to collect when the analysis is retrospective. This problem of uncertain data is partly solved by the revised version of the Treasury Board Benefit-Cost Analysis Guide (Watson & Mallory, 1997), which allows probabilistic estimates of program results to be used in the analysis. There are not yet many examples of this technique in practice. One is Transport Canada's evaluation of alternative requirements for small commercial vessels to carry emergency signaling equipment. This article describes that evaluation and assesses how well the methodology worked.