OBJECTIVES: Probabilistic sensitivity analysis (PSA) of economic evaluations has become more important because decision makers want to know how uncertain results are. We evaluated the consistency of PSA methods and how well they adhered to published recommendations. METHODS: Publications of methodological guidelines on the conduct of PSA were reviewed by three researchers, and a consensus set of criteria for assessing PSAs were developed. English language health economic evaluations based on models, published in 2006 were obtained from a PubMed search using terms “probabilistic sensitivity” or “probabilistic uncertainty”. Methodological items were extracted from each article and independently evaluated against the criteria by each researcher. Disagreements between evaluations were resolved through voting by the entire research team. RESULTS: All 43 economic evaluations identified by the search criteria were reviewed. 86% varied probability and utility inputs but only a minority (25%) did so for cost and resource use inputs. 79% of the studies reported parameter ranges used in the PSA but only half provided rationale for the ranges selected. The majority of analyses (65%) used a single data source to inform distributions, rather than attempt to integrate findings from multiple studies. Parameter correlation was only addressed in one instance and only two studies incorporated structural uncertainty in their analyses. In half of the studies, PSA was the only type of sensitivity analysis conducted, with no one- or multi-way sensitivity analyses. Cost-effectiveness acceptability curves derived from the PSAs were presented in all cases. Less than 10% of studies discussed limitations of their PSA. CONCLUSION: Although PSA has been pushed as standard practice for economic evaluations, the quality of these analyses was mixed. Greater consistency in terms of inclusion of inputs varied and more transparency in describing development of input probability distributions in the conduct of PSAs should improve quality and cross-study comparability of results.