There are known difficulties in measuring testosterone in women using 'direct' (non-extraction) immunoassays. The positive bias encountered in many of these assays has been attributed to an unidentified interfering substance. We have evaluated the potential for dehydroepiandrosterone sulphate (DHEA-S) to cross-react in the Abbott Architect testosterone assay. Samples (n = 81) from female patients who had DHEA-S, direct and extracted testosterone measured were examined retrospectively. Possible cross-reactivity was evaluated by spiking both testosterone free assay diluent and serum from a woman with exogenous DHEA-S. The retrospective patient analysis of the data demonstrated a relationship between interference and DHEA-S (r = 0.78, P < or = 0.0001). Spiking studies revealed a linear relationship between added DHEA-S and the measured testosterone concentration. Consistent with the retrospective study, for every 1 micromol/L increase in DHEA-S, 0.23 nmol/L of testosterone was measured in the diluent. Spiking DHEA-S into serum gave only a 0.12 nmol/L increase in measured testosterone for every 1 micromol/L rise in DHEA-S. DHEA-S cross-reacts in the Abbott Architect direct immunoassay for testosterone to a clinically significant effect. This cross-reactivity is likely to be responsible for much of the interference observed when measuring testosterone in women by this method.