Less than 20 percent of elderly and other high-risk persons targeted for annual influenza vaccination are immunized each year. In most busy practice settings, it is difficult for primary care physicians to identify every patient in need of preventive health interventions. The purpose of this study was to assess the effect of microcomputer-generated reminders on influenza vaccination rates in a university-based family practice center. The practice uses an interactive encounter form system from which updated clinical information is routinely entered into a cumulative database. During a 2-month period, 686 patients were identified in the database as eligible to receive influenza vaccine according to accepted criteria. Practice physicians (n = 32) were stratified by level of training and randomized to one of three groups, thereby receiving printed reminders on the encounter forms of all, none, or half of their eligible patients. Patients of physicians who always received reminders were more likely to receive influenza vaccine during the study period than patients of the never-reminded physicians (51 percent versus 30 percent, P less than 0.001). Patients whose physicians received reminders for only half their patients had an intermediate likelihood of receiving a vaccination if a reminder was printed (38 percent) but were less likely than the patients of never-reminded physicians to receive the vaccine if no reminder was printed (20 percent, P less than 0.001). This study suggests that physicians learn to depend on reminders for preventive health activities and that reminders are most effective when they are provided at every patient encounter.