Human proinsulin and 32–33 split proinsulin have been measured in the peripheral circulation by immunoradiometric assays (IRMAs) and have been shown to be elevated in impaired glucose tolerance and non-insulin-dependent diabetes mellitus (NIDDM). The IRMA for 32–33 split proinsulin did not discriminate between this molecule and des-32 or des-31,32 split proinsulin. We describe the comparison of IRMA for human plasma proinsulin and 32–33 split proinsulins with assays combined with high-performance liquid chromatography (HPLC), which can discriminate between 32–33 split, des-32 split, and des-31,32 split proinsulin. Subjects were those with normal glucose tolerance ( n = 8) and those with NIDDM ( n = 17), who were studied while fasting and 30 min after a glucose load. After collection, blood was centrifuged promptly, and the serum/plasma was stored frozen until assay. Both IRMA and HPLC methods were calibrated against synthetic peptides. Interassay coefficients of variation for the IRMA for proinsulin and 32-33 split proinsulin were < 13% over the ranges 3.8–65 pmol/l and 6.4–65 pmol/l, respectively. The following regression lines were obtained: proinsulin IRMA ½ −0.143 + 1.066 HPLC, r = 0.860; 32–33 split proinsulin IRMA ½ 0.048 + 1.051 HPLC; and des-31,32 split proinsulin, r = 0.814. For both analytes, there was no significant difference in the relationship of IRMA to HPLC results between the various subject groups and various time points. Thus, the IRMA for proinsulin has been validated by an independent method. The 32–33 split proinsulin IRMA predominately measures des-31,32 split proinsulin, which is the major partially processed proinsulin present in human plasma of both normal and NIDDM subjects.