Several years ago we developed a novel two-site immunoradiometric assay (IRMA) for dimeric inhibin. However, relative to the purified 32 kDa bovine inhibin standard used at that time, the immunopotencies of crude inhibin-containing samples were much less than their biopotencies estimated by pituitary cell bioassay. In attempting to improve assay performance and resolve this discrepancy we recently discovered that introduction of a preassay oxidation step to the IRMA results in a dramatic increase in the immunopotencies of inhibin-containing test samples (e.g.: bovine, human, porcine follicular fluid (FF)) and of a new (purified in 1993) 32 kDa bovine inhibin standard. However, the oxidation step did not affect the immunopotency of our original standard (purified in 1987), indicating that this material had undergone spontaneous oxidation during long-term storage, thus accounting for its higher immunopotency in our original IRMA and providing an explanation for the discrepancy between immunoactivity and bioactivity referred to above. These findings, together with other observations on the behaviour of oxidized and non-oxidized samples of inhibin, related peptide fragments and inhibin-containing samples in the IRMA and α subunit radioimmunoassay (RIA), indicate that the anti-βA 82–114 monoclonal antibody (E4) used as tracer in the IRMA binds selectively to the oxidized (Met O89,91,108) form of the peptide. This property of the antibody can be exploited to advantage by incorporating simple modifications to existing inhibin/activin immunoassays to ensure that all samples and standards are fully oxidized before antibody addition. Dimeric inhibin levels in individual bovine FF samples (n=105) estimated by the 'oxidized' IRMA (9·66 ± 0·50 ng/μl) were highly correlated (r=0·90; Pr=0·89; PJournal of Endocrinology (1994) 141, 417–425