When characterization of the specificity of an antiserum for radioimmunoassay (RIA) is performed by the conventional method, the conditions under which interference occurs are not respected because of the lack of specific antigen. We have studied the behavior of antisera reproducing the real environment existing in unknown samples, in which antigen, interferent and tracer complete simultaneously. A testosterone (T) antiserum and a 5 alpha-dihydrotestosterone (D) anti serum were characterized by setting up two distinct hapten recovery tests in the presence of both the hapten and the crossreactant added to steroid-free serum in various concentrations in order to reproduce multiple concentration ratios. These samples, together with the standard curves samples (prepared by 'spiking' steroid-free serum with known concentrations of T or D) were extracted and subjected to T-RIA and D-RIA without purification. The results have shown that the interferent-induced incremental ratio is a linear function of the ratio of the levels of cross-reactant and hapten via a proportionality factor inversely correlated to the antiserum specificity. By means of this function, the overestimated T and D levels found in samples after 'extraction only' have been corrected and the resulting values have shown acceptable correlation with the corresponding levels determined after column chromatography.