A routine radioimmunoassay for human proinsulin in serum has been developed. The reagents used were: antibodies against the C-peptide part of the proinsulin molecule, human proinsulin as the standard and 125I-labelled synthetic human Tyr-C-peptide as the tracer. The first step in this assay comprises the binding of proinsulin to insulin antibodies covalently coupled to Sepharose (S-AIS). Although bound to the solid-phase S-AIS, the proinsulin retains its second immunogenic site, viz., the C-peptide part of the molecule, accessible. Hence a surplus of C-peptide antibodies is added to the S-AIS-bound proinsulin, and the residual amount of C-peptide antibody is determined by addition of 125I-Tyr-C-peptide. The detection limit is approximately 0.01 pmol/ml. The advantages of this method are: (1) its high specificity (proinsulin is determined as a molecule having both an insulin and a C-peptide moiety), (2) its simplicity and rapid performance, and (3) the low detection limit of the assay. Fasting sera from 24 nondiabetics, 9 maturityonset diabetics and 10 newly diagnosed insulin requiring diabetics showed the following concentrations of proinsulin: 0.009 _4-0.005, 0.022 + 0.23 and 0.010 _+ 0.009 pmol/ml (mean + SD). One hour after 1.75 g/kg oral glucose, the values increased to 0.052 _+ 0.023, 0.046 + 0.022 and 0.032 + 0.022 pmol/ml. The fasting proinsulin constituted 19, 23 and 17% of the IRI, respectively, whereas 1 h post glucose these values changed to 8, 9 and 31% of IRI. Serum from 10 insulin-treated diabetics containing insulin antibodies contained from 0-1.80 pmol/ml, whereas the C-peptide levels in the same patients were 0-0.35 pmol/ml. It is suggested that insulin requiring diabetics hypersecrete proinsulin due to the inability of their B-cell to arrange proinsulin in secretory granules for adequate proinsulin/ insulin conversion.