The aim of our study was to investigate the efficacy of prednisone to preserve pancreatic beta-cell function in patients with recent-onset Type I diabetes mellitus (IDDM). Twenty-five patients with IDDM, aged 24 +/- 6 years, entered the trial within 8 weeks of the onset of diabetes. They were allocated, according to a single blind randomized protocol, to one of the following treatments: (A) prednisone (15 mg/day), (B) indomethacin (100 mg/day), (C) placebo. All treatments lasted 8 months and all patients achieved satisfactory metabolic control with a multi-injection regimen (three injections/day) within a few weeks, and maintained it throughout the entire period of observation. Only minor side effects were observed in the prednisone-treated patients. A lower insulin requirement was observed in the prednisone group than in other patients at 12 months (0.33 +/- 0.11 vs 0.57 +/- 0.06 U/kg/day, P less than 0.05), 18 months (0.34 +/- 0.11 vs 0.64 +/- 0.06, P less than 0.05) and 24 months (0.38 +/- 0.10 vs 0.63 +/- 0.05, P less than 0.05). Endogenous insulin release, evaluated as urinary C-peptide, was higher in the prednisone group than in other patients at 3, 6, 9, 12, 18 and 24 months (P less than 0.05). ANOVA confirmed differences among the three groups. Our study indicates that prednisone administration, at low doses and for a long period of time, effectively restored endogenous insulin release in IDDM patients.