OBJECTIVE: To address two specific issues relating to the conduct of economic analysis based on decision analytic techniques:- the impact on results of different sources of both effectiveness and resource use data. METHODS: Sensitivity analysis of erythropoietin use in orthopedic surgery and alendronate in the treatment of osteoporosis. Analysis focuses on the effect on the incremental cost-effectiveness ratio of different sources of treatment effectiveness and resource use. For effectiveness, analysis compared results based on individual clinical trials and meta analysis. Further analysis compared the results based on resource use data from the clinical trials and data from administrative data sets. RESULTS: Results varied greatly between the different sources of effectiveness and resource use data. Based on outcomes from individual clinical trials the cost per QALY gained from alendronate varied from $27000 to $104500. The cost per QALY gained from a meta analysis of all studies was $35600. The cost per life year gained of erythropoietin was more than ten times greater using actual resource use data in clinical practice in Ottawa than using trial based data. CONCLUSIONS: Economic analysis conducted through decision analytic techniques has been open to criticism over the ease at which desired results can be obtained. By adopting standards based on an evidence-based approach to identifying parameters, such criticisms can be avoided. Based on the results of this paper we recommend that meta analysis is the preferred source of outcome data but routine local data are the best source of resource use data.