The effect of corticosteroids on beta cell function and humoral immune response in type 1 diabetes was tested in a 2-month trial conducted on 32 newly diagnosed patients (age 22.8 +/- 1.4 years, mean +/- S.E.M.). Prednisone was administered at immunosuppressive dosage (1 mg.kg-1.day-1) during the initial 10 days and at a maintenance dosage (0.3 mg.kg-1.day-1) for 50 days. Patients (n = 32) were enrolled within 6 weeks after diagnosis and matched in pairs for age, sex, presence of islet cell antibodies (ICA) and glucagon stimulated C-peptide levels. Insulin discontinuation was not contemplated. All the patients who received prednisone became ICA during treatment but in some (4 out of 10) this effect was only transient. Insulin antibodies (IA) were significantly lower in the prednisone group at second and third month (P < 0.05). No patient experienced complete remission but in 10 prednisone and 4 control patients the insulin requirements were below 0.3 IU/kg (P < 0.05). With similar glycemia the fasting C-peptide levels were higher in the treated patients. The profile of the insulin requirements during the follow-up was different in the two groups and at 9 months the prednisone group needed less insulin than the control (P < 0.05). Interestingly, within the prednisone-treated group and after 6 months, the levels of stimulated C-peptide improved significantly among the ICA+ patients while they were steady or declined in ICA- (P < 0.01). The analysis of variance covariance confirmed a positive interaction between ICA and the administration of prednisone on the outcome of beta cell function.(ABSTRACT TRUNCATED AT 250 WORDS)