In individual serum samples, the sex hormone-binding globulin (SHBG) binding capacity for dihydrotestosterone (DHT) was systematically found to be decreased by 30-60% when either gelatin or dithiothreitol (DTT) was present in the assay buffer. The presence of gelatin in the buffer prevented DTT from further decreasing the SHBG binding capacity of serum samples, suggesting a similar mechanism of action on SHBG for both of these substances. This observation led us to compare the molecular forms of SHBG by high performance liquid chromatography on a TSK G 3000 SW column, in the presence or absence of DTT. When undiluted serum previously incubated with [3H]DHT was chromatographed, only monomeric SHBG could be detected, independently of the presence or absence of DTT in the elution buffer. When the serum was diluted, incubated and chromatographed with buffer devoid of DTT, a dimeric SHBG peak was progressively observed, as a function of the sample dilution. Furthermore, for a given serum dilution, the relative size of the dimeric SHBG peak was also dependent on the steroid concentration present in the sample. By contrast, when serum was diluted, incubated and chromatographed with DTT-supplemented buffer, only the SHBG dimer peak could be detected. These results suggest that in serum, in vitro at least, SHBG is present in its monomeric form. Serum dilution with buffer devoid of DTT or gelatin induces the progressive dimerization of the protein, resulting in a progressive decrease of its apparent binding capacity. This could explain the great discrepancies of SHBG levels as reported in the literature. Because serum dilution with buffer supplemented with DTT or gelatin induces the complete dimerization of SHBG, independently of the sample dilution, we suggest that these substances be routinely used for the measurement of SHBG binding capacity. The SHBG binding capacity obtained in these latter conditions reflects however half the binding capacity of undiluted serum.