OBJECTIVES: In economic evaluations, failure to use the best available evidence for the estimates of clinical effect has attracted little attention, but may lead to biases in study results. This is a particular concern in the evaluation of pharmaceuticals, given the industrial sponsorship in many studies. Therefore, the objective was to investigate whether economic evaluations of drug therapies use the best available clinical effectiveness evidence, appropriate to the population likely to receive the drug. METHODS: A random sample of 50 economic evaluations of drug therapies published in 2001/2002 was selected from the NHS Economic Evaluation Database. Study design, reporting quality and external validity of the clinical data were assessed using methodological checklists. To judge the applicability of the effectiveness data, participants and settings were compared with standard clinical practice. For each therapy, alternative, high quality sources of clinical effectiveness data were sought in the Cochrane databases. RESULTS: Approximately half of the studies utilised estimates of clinical effect from randomised controlled trials. Quality of reporting was good, but measures of external validity were uniformly poor. There were potential problems with the applicability of half the studies, in particular limited data on settings, use of specialised centres and inappropriate gender balance. Relevant, potentially usable systematic reviews were identified for 27 (54%) of the economic evaluations in the sample, although none had been used. The use of data from systematic reviews would change the size of reported cost-effectiveness ratios. CONCLUSIONS: These findings suggest that critical appraisal of the clinical data underlying economic evaluations is needed before they are used in health care decision-making. Best available clinical evidence is not being utilised and further research is indicated to quantify the implications of using poor quality, or selective effectiveness data in economic evaluations.