This multicenter study of patients with non-insulin-dependent diabetes mellitus (NIDDM) was undertaken (1) to determine the incidence of insulin antibody formation in such patients before exposure to exogenous insulin; (2) to assess the long-term immunologic response to semisynthetic human insulin (ssHI) in new insulin users and in patients transferred from animal insulin; and (3) to document the efficacy and safety of ssHI in both new and transfer patients. In addition, a substudy at one participating center was designed to compare the effects of a twice-daily versus a once-daily regimen in initiating ssHI therapy in new patients with uncontrolled NIDDM. Among the 37 new patients, only one had detectable insulin antibody levels before administration of insulin. After ssHI therapy was begun, this patient's antibody levels rapidly fell below the assay's limit of detection. Detectable levels of antibodies to human insulin were found in only 36% of 28 new patients after 12 months of therapy. As expected, the prevalence of insulin antibodies among animal-insulin users was high: 82% of the 17 transfer patients had detectable insulin antibody levels (mean, 2.27 mU/ml) at baseline. After six months of treatment with ssHI, antibody levels decreased significantly (mean, 0.75 mU/ml; P less than 0.05). Control of glycemia was assessed by measurement of glycosylated hemoglobin. Values decreased significantly (P less than 0.01) in the new patients after the introduction of ssHI and remained stable in the transfer group after initiation of ssHI therapy. Hypoglycemic episodes were infrequent in both groups. In initiating ssHI therapy in new patients hospitalized with uncontrolled NIDDM, a twice-daily regimen resulted in a more rapid normalization of glycemia and earlier discharge than did the standard once-daily regimen. In conclusion, the results of this study provide further evidence that NIDDM and insulin-dependent diabetes mellitus (IDDM) are immunologically different disorders, with the immune system probably not involved in the pathogenesis of NIDDM. The data also indicate that ssHI is less immunogenic than animal insulin and that it is effective and safe in the management of NIDDM both in first-time insulin users and in patients transferred from animal-species insulin. Thus ssHI would appear to be useful in treating NIDDM, especially in patients who require intermittent insulin therapy.