An intervention package was examined to determine its effectiveness in increasing office visits and in reducing the incidence of nonelective hospitalizations (those for urgent or emergent reasons). The intervention included mailings of information, appointment reminders, and intense follow-up by telephone of visit failures for rescheduling. Eight hundred fifty-four patients receiving drug therapy for diabetes mellitus were stratified by risk of nonelective hospitalization and randomly assigned to the control group or the intervention group. After two years, the intervention group averaged 9.1 per cent more kept scheduled visits per month than the control group (0.371 vs. 0.340, p = 0.02). However, the mean incidence of nonelective hospitalizations per month was not significantly different between intervention and control groups (0.040 vs. 0.041, p = 0.9), nor was there a difference in nonelective hospital days per month (0.443 vs. 0.425, p = 0.7). The authors conclude that while mailings and telephone calls can increase office visits, the intervention is not sufficient to reduce morbidity necessitating nonelective hospitalizations of diabetic patients.