To study the effects of computer-generated reminders for influenza vaccination of patients at high risk of pulmonary disease during the winter, we selected 4555 such patients from a population of 15,000 adults who participated in a three-year randomized trial of preventive-care reminders. The physicians who received the reminders vaccinated eligible patients twice as often as did the control physicians (P = 0.0001). There was a linear increase in the incidence of influenza in our area during the three winters under study (from 1000 to 33,451 to 71,075 cases per year), and we modeled the percentage of patients with hospitalizations, emergency room visits with chest radiography, and blood gas determinations as a logistic function of this increase. The difference in linear trends between the patients in the intervention group (whose physicians received reminders) and those in the control group was significant for emergency room visits (P less than 0.05), hospitalizations (P less than 0.01), and blood gas determinations (P less than 0.001). The most likely explanation for the difference is the greater use of influenza vaccine in the intervention group.