To obtain the confidence interval (CI) of the results of a cost-effectiveness model for the treatment of hypercholesterolemia. Background: Incremental CE ratios play an important role in pharmacoeconomic analysis. However, CI estimates for the CE ratio are rare because the underlying distribution is unknown in most cases. Rather than using traditional one-way or multi-way sensitivity analysis to explore estimate variability, a non-parametric bootstrap method is used to generate the CE ratio CI. This is more robust because it does not make parametric assumptions. METHODS: 1000 bootstrap samples were drawn simultaneously with replacement from both an empirical distribution and a Monte Carlo experiment. For patient risk factors we used the complete subsample of men age 45-64 fitting NCEP primary prevention guidelines in the National Health and Nutrition Examination Survey III. For model parameters (e.g., costs, discount rate) values were randomly drawn from a specified multivariate distribution. With 1000 repeated estimates, the 26th and 975th ranked CE ratios were used to define the 95% confidence interval. Programming was done in Visual Basic for Applications (VBA) language. The study perspective was that of a third party payer. Efficacy and compliance data were obtained from the WOSCOPS clinical trial. The incremental 5-year cardiovascular health care costs were derived from a health insurer database. RESULTS: A 95% CI for the CE ratio resulting from the nonparametric bootstrap method showed that primary prevention of cardiovascular disease with pravastatin remained cost-effective. CONCLUSION: Confidence interval estimation for the CE ratio is important and will allow individuals to identify if a new intervention is a good value for cost, while capturing the inherent imprecision in the underlying CE model parameter values.