Testosterone (T) administration slows LH pulse frequency in man, presumably by an effect on the hypothalamic GnRH pulse generator, but it also may have a direct action on the pituitary. To determine if T does indeed affect gonadotropin secretion by acting directly on the pituitary, we studied the effect of T on GnRH-stimulated gonadotropin secretion. Six men with hypogonadotropic hypogonadism were treated with physiological doses of GnRH (5 micrograms every 2 h, sc by automatic infusion pump) for 6 weeks. Once their gonadotropin levels were normal, the men received a supraphysiological dosage of T enanthate (200 mg, im, weekly for 8 weeks) in addition to GnRH. They then received GnRH alone for a final 8-week period. Blood sampling was performed every 10 min for 8 h at the end of each of the three study periods. T administration suppressed the mean serum LH level to about 50% of the value during GnRH alone [18 +/- 2 (+/- SE) vs. 37 +/- 2 micrograms/L; P less than 0.05] and suppressed the mean serum FSH level to about 30% of the value during GnRH alone (39 +/- 6 vs. 128 +/- 28 micrograms/L; P less than 0.05). Eight weeks after stopping T, while continuing GnRH alone, serum LH and FSH levels were similar to those at the end of the first period of GnRH administration. The mean LH response to GnRH was reduced during T administration (17 +/- 3 micrograms/L) compared to that during the initial period of GnRH alone (31 +/- 4 micrograms/L; P less than 0.05). Serum T and estradiol levels were in the low normal range after GnRH alone before T administration (11 +/- 2 nmol/L and 105 +/- 17 pmol/L, respectively) and increased to just above the normal adult ranges after 8 weeks of T administration (36 +/- 5 nmol/L and 264 +/- 49 pmol/L, respectively). These results demonstrate that T and/or its metabolites inhibit LH and FSH secretion by a GnRH-independent mechanism, probably directly on the pituitary gland, in man.