In economic evaluations of health treatments, the sensitivity of a cost-benefit (CB), cost-effectiveness (CE) or cost-utility (CU) analysis to changes in modeling assumptions, variation in data, and sampling error is important. The typical approach to this problem is ad hoc experimentation; namely, a few parameters of particular interest are changed, either separately or in combination, over plausible ranges. The impact of random variation in the data is seldom explored beyond parametric tests of the statistical significance of estimated coefficients. This note suggests a systematic approach to sensitivity analysis. Bootstrap sampling is used to determine to what extent the patients' response to treatment and economic consequences might vary due to many replications of a clinical trial.