RIA methodology is used widely to measure proinsulin in human serum. However, some RIAs lack the sensitivity necessary to quantify proinsulin in unextracted serum and require long incubation periods. We developed an RIA with a sensitivity of 3.5 pM that permits the routine measurement of proinsulin in less than 48 h. This was accomplished by using a nonequilibrium binding reaction at room temperature and PEG-assisted second antibody precipitation as the method for separating bound and free proinsulin. We obtained a specific antiproinsulin antibody by absorbing the initial goat antiserum with human C-peptide-agarose. Proinsulin produced 50% displacement of tracer at 25.6 pM, whereas both human insulin and C-peptide failed to displace tracer at concentrations as high as 1 microM. We evaluated several cleaved derivatives of proinsulin for cross-reactivity with the antibody. B-chain-C-peptide cleaved derivatives (less than or equal to 50% cross-reactivity) were more potent than A-chain-C-peptide cleaved derivatives (less than 5% cross-reactivity). However, all derivatives cleaved in the region from 56-60 failed to cross-react with the antiserum. These data indicate that a major antigenic determinant is present on the C-peptide region of proinsulin adjacent to the A-chain-C-peptide junction. After administration of an oral glycemic challenge, the mean fasting serum concentration of proinsulin in normal adults rose from 4.1 +/- 0.28 to 23.6 +/- 3.8 pM. We found a significant difference in the proinsulin concentrations in 6 adults before and after a glycemic challenge when two different antibodies were used in the RIA.(ABSTRACT TRUNCATED AT 250 WORDS)