We investigated the effects of long term testosterone (T) administration on pulsatile gonadotropin secretion in agonadal women and the effects of estradiol (E2) on gonadotropin secretion in eugonadal women in the follicular phase of the menstrual cycle. We studied 4 groups: A) 28 eugonadal women in the early follicular phase of the menstrual cycle, B) 11 hypogonadal women, C) 13 agonadal female to male (f-t-m) transsexuals treated for at least 3 months with 120-160 mg T undecanoate (TU)/day, orally, and D) 5 agonadal f-to-m transsexuals treated for at least 6 months with 250 mg of a mixture of testosterone esters, im (im T-esters), every 2 weeks. The eugonadal women in the early follicular phase had a mean serum E2 level of 193 ± 94 (±SD) pmol/L, significantly higher (P < 0.01) than that in the hypogonadal women (60 ± 24 pmol/L), whereas there was no difference in the mean serum T levels (1.8 ± 0.7 vs. 2.3 ± 1.5 nmol/L). The higher serum E2 level in the eugonadal women was associated with a significantly lower mean serum LH level (6.9 ± 2.6 vs. 44.6 ± 17.6 U/L; P < 0.01) and LH pulse amplitude (2.8 ± 1.0 vs. 12.6 ± 4.8 U/L; P < 0.01), whereas the mean nadir LH interval did not differ between the two groups (75 ± 29 vs. 81 ± 49 min). The mean serum T level in the agonadal f-to-m transsexuals treated with oral TU was significantly higher (P < 0.01) than that in the hypogonadal women (9.7 ± 4.7 vs. 2.3 ± 1.5 nmol/L). In spite of this elevated T level there was no difference in the mean serum LH level (38.4 ± 14.7 vs. 44.6 ± 17.6 U/L), LH pulse amplitude (14.3 ± 5.7 vs. 12.6 ± 4.8 U/L), or nadir LH interval (72 ± 27 vs. 81 ± 49 min) in these groups. Also, the mean serum E2 (64 ± 16 vs. 60 ± 24 pmol/L and FSH levels (62 ± 17 vs. 64 ± 28 U/L) did not differ between these groups. Treatment of the agonadal f-to-m transsexuals with im T-esters resulted in mean serum T and E2 levels of 34.4 ± 27.0 nmol/L and 121 ± 54 pmol/L, respectively, both significantly higher (P < 0.01) than those in groups B and C. Their mean serum LH (19.3 ± 6.4 U/L) and FSH (32 ± 10 U/L) levels were significantly lower (P < 0.01) than those in groups B and C; the mean nadir LH interval was significantly longer (131 ± 55 min; P < 0.01), whereas the LH pulse amplitude (8.8 ± 1.7 U/L) was similar. From these results we conclude that 1) the serum T levels in the agonadal f-to-m transsexuals treated with oral TU (9.7 ± 4.7 nmol/L) did not have an effect on gonadotropin secretion, whereas those in the agonadal f-to-m transsexuals treated with im T-esters (34.4 ± 27.0 nmol/L) decreased the mean serum LH level, LH pulse frequency, and FSH level; 2) the E2 levels attained in the early follicular phase did not have a suppressive effect on the LHRH pulse generator, since the LH pulse frequency did not differ between the hypogonadal women and the eugonadal women in the early follicular phase, despite the significantly higher mean E2 level in the latter group; 3) the suppressive effects of high T levels on LH pulse frequency were attributable to androgen action rather than to androgen-derived estrogen action, since serum E2 values were similar to those in the eugonadal women; 4) androgen levels in these experiments in T-treated agonadal female subjects were comparable to or exceeded those in women with polycystic ovarian disease, yet the agonadal females did not have abnormalities in gonadotropin secretion characteristic of polycystic ovarian disease.