Testosterone at physiological levels cannot exert negative feedback action on LH secretion in long-term castrated male monkeys. The cellular basis of this refractoriness is unknown. To study it, we compared two groups of male rhesus macaques: one group (group 1, n = 4) was castrated and immediately treated with testosterone for 30 days; the second group (group 2, n = 4) was castrated and treated with testosterone for 9 days beginning 21 days after castration. Feedback control of LH by testosterone in group 1 was normal, whereas insensitivity to its action was found in group 2. Using the endpoints of concentrations of aromatase activity (P450_AROM messenger RNA [mRNA]) and androgen receptor mRNA in the medial preoptic anterior hypothalamus and in the medial basal hypothalamus, we found that aromatase activity in both of these tissues was significantly lower, P < 0.01, in group 2 compared with group 1 males. P450_AROM mRNA and androgen receptor mRNA did not differ, however. Our data suggest that the cellular basis of testosterone insensitivity after long-term castration may reside in the reduced capacity of specific brain areas to aromatize testosterone. Because P450_AROM mRNA did not change in group 2 males, we hypothesize that an estrogen-dependent neural deficit, not involving the regulation of the P450_AROM mRNA, occurs in long-term castrated monkeys.