We investigated the immunogenicity of human proinsulin (HPI) when used as the sole or principal insulin agonist in insulin-naive patients with insulin-dependent (type I) and non-insulin-dependent (type II) diabetes mellitus. Sixty-one patients (13 type I, 48 type II) were treated with rDNA human insulin (NPH HI with or without regular HI) and 53 were treated with HPI (8 type I, 45 type II). At 6 mo, virtually identical levels of HbA1c (5.2 vs. 5.3%, P = NS) were achieved. However, regular HI was added less often to the treatment regimen in HPI-treated patients (16 vs. 32 patients, P < .001). Overall, there was no significant increase in proinsulin-specific antibodies in either treatment group. However, 8 of 51 (1 transiently) patients in the HPI group developed low levels of binding of HPI (highest percentage bound was 5%). Two patients in the HI group developed very low levels of HPI binding (1.2 and 1.9%). Binding of HI (> 2.4%) was seen in both treatment groups; however, the prevalence of HI binding was less in the HPI group at 6 mo (39 of 60 in HI group vs. 20 of 51 in HPI group, P = .008). Concomitant treatment with regular HI did not affect the prevalence or level of binding of HPI or HI. We conclude that human proinsulin is a weak immunogen when used as the principal insulin agonist and may reduce both the formation of anti-HI antibodies and the need for concomitant therapy with regular HI.