The purpose of this study was to compare the proportion of therapeutic interventions that is supported by scientific evidence between hospital-based and community-based primary care clinics. The study setting is a primary care clinic in university-affiliated tertiary hospital and community-based primary care clinic in a region in Seoul. A retrospective review of patients' medical records was done according to the primary diagnosis and treatment for 890 patients in community-based clinic and 307 in hospital-based clinic. The evidence for the intervention was searched for in Medline (1966-2001) and other evidence-based databases. The evidence was then classified as one of the three categories (i) evidence from randomised controlled trials (RCTs); (ii) convincing non-experimental evidence; and (iii) interventions without substantial evidence. In the hospital-based clinics, out of 356 primary diagnosis and treatment pairs, 59.6% were supported by evidence from at least one RCTs, and 19.4% were supported by convincing non-experimental evidence. In the community-based clinics, 47.1% out of 893 pairs and 10.5% fell into each category. As a result, the proportion of interventions in community settings that were based on evidence meeting our criteria was significantly lower than that of the hospital-based settings. To facilitate evidence-based practice, supportive activity should be focused on the primary care settings based on community where medical practices are less evidence-based.