To test the hypothesis that physician education is an effective strategy to reduce total hospital costs, we evaluated three educational interventions at a large university hospital. This prospective controlled study spanned two academic years and involved 1,663 patients and 226 house staff. In the first year, weekly lectures on cost containment (medicine and surgery) and audit with feedback (medicine only) both failed to produce a significant change in total hospital charges. The "dose" of the intervention was increased on medicine in the second year by combining the lecture and audit strategies. Again, total charges did not change significantly. While decreased use occurred for certain selected services, the impact was not great enough to affect total hospital charges significantly. We conclude that, in the absence of other cost containing incentives, physician education alone is not an effective hospital cost containment strategy.