Ketoconazole inhibits testosterone biosynthesis in men, but the exact site of its action on the androgen pathway remains to be established. To examine this question, we measured several steroids in the androgen and glucocorticoid pathways in normal men before and after receiving either a single dose of 200 mg ketoconazole or placebo in a cross-over randomized trial. Total and free plasma testosterone fell to levels 60% below basal within 4-8 h (P less than 0.02 in all) and then returned to control concentrations by 24 h after drug administration. The transient alterations of plasma testosterone correlated well with ketoconazole blood levels, which peaked at 2 h and fell exponentially thereafter. A compensatory increase in plasma LH at 24 h in the drug but not placebo group was consistent with the decrease in plasma testosterone. The levels of plasma androstenedione paralleled those of testosterone in the ketoconazole-treated subjects. In marked contrast, plasma 17 alpha-hydroxyprogesterone increased at 4-8 h (all P less than 0.02) before returning to basal values at 24 h. This rise in precursor with fall in product steroid implicated an effect of ketoconazole on the C17-20 lyase enzyme. This conclusion was supported by the highly significant increase in the ratio of plasma 17 alpha-hydroxyprogesterone to androstenedione observed between 2 and 24 h after drug administration. The effect of ketoconazole at this dose level appeared relatively specific, since no decrements in plasma cortisol or 11-desoxycortisol were found. During chronic administration of 200 mg ketoconazole daily, decrements of plasma testosterone 2-4 h after drug administration were minimal and documented only by paired comparisons within subjects but not by unpaired tests between normal men and men receiving drug. The lack of major effects on testosterone levels long term at this dosage probably explain why few androgen-related side effects with this drug were previously reported. Ketoconazole, therefore, represents another compound with relatively selective effects on a cytochrome P-450-mediated steroid hydroxylation step, namely that involved with C17-20 lyase.