OBJECTIVES: Guidelines for conducting cost-utility analyses (CUAs) contain inconsistent recommendations for selecting cost, quality of life, and discount rate parameters. Sensitivity analyses can indicate whether adhering to different guidelines results in different policy recommendations. The purpose of this study is to investigate the use of sensitivity analyses to test economic parameters in the cost-utility literature. METHODS: Recommendations from published guidelines are summarized. CUAs of pharmaceutical therapies identified in a prior study (N=71 articles) were reviewed and further audited. We identified threshold CU ratios (N=36) and base cases for which sensitivity analyses were reported (N=123). For each base case, up to 2 sensitivity analyses for cost (N=97), quality of life (N=136), and discount rate (N=127) were examined. RESULTS: There are substantial disagreements among the guidelines regarding economic parameters. The most frequently mentioned threshold CU ratios were $20,000/QALY, $50,000/QALY, and $100,000/QALY. The proportions of sensitivity analyses reporting quantitative results that crossed the threshold above the base case CU ratio were 23% for cost sensitivity analyses, 38% for quality-of-life sensitivity analyses, and 15% for discount rate sensitivity analyses. There was no difference in quality ratings between CUAs that reported sensitivity analysis results that exceeded the thresholds (N=17) and those that did not, but the overall quality and completeness ratings were only moderate. CONCLUSIONS: Sensitivity analyses for economic parameters are widely reported and can be used to identify whether choosing different assumptions leads to a different decision. Different decisions occur more frequently for cost and quality-of-life assumptions than for discount rate assumptions. Sensitivity analyses for cost and quality-of-life parameters should be used to test alternative guideline recommendations, but sensitivity analyses for discount rates do not have the same import. Adhering to recommendations on performing cost-effectiveness analyses would improve the overall quality of these types of studies.