Prophylactic insulin treatment is effective in preventing diabetes in animal models of insulin-dependent diabetes mellitus (IDDM) but the safety of such preventive treatment in prediabetic human subjects remains unclear; insulin is a potential autoantigen that could accelerate beta-cell decompensation and onset of IDDM. We have investigated whether insulin treatment of non-diabetic subjects increases the risk of subsequent development of diabetes in a retrospective study of Danish patients who received insulin-shock treatment for psychiatric disorders. Mean age of the 481 patients at insulin-shock treatment was 32.6 (range 12.9-69.6) years. The patients received 59 (6-200) injections of 78 (16-261) IU bovine/porcine insulin. Hospital records provided an average of 22.0 (0.6-51.2) years' observation. During the observation time, IDDM developed in only 1 patient; 1.3 cases would be expected from Danish incidence data (p = 0.75). Similarly, there was no significant difference between the observed number of cases of non-insulin-dependent diabetes mellitus (NIDDM) and the number expected from Danish prevalence data (12 vs 10.2; p = 0.45). We collected blood samples from 27 of the patients. All but 2 (who had previously diagnosed NIDDM) had normal fasting blood glucose and plasma insulin concentrations, none had islet-cell antibodies, and only 2 had detectable insulin antibodies. Thus, the risk of diabetes was not increased by the use of many insulin injections in these non-diabetic subjects. We conclude that clinical trials on prevention of IDDM by prophylactic insulin treatment can be regarded as safe.