We have analyzed the impact of a clinical information system designed to give feedback on profiles of medical practice in two hospitals. The system was planned to use group influence to induce change among deviant physicians. The information system failed to modify the physician's practice profiles. A large part of this failure is linked to the small variation recorded in utilization patterns among physicians. Medical practice appears to be quite homogeneous in a setting such as a hospital where the clinical behavior of physicians is highly visible to colleagues.