In an attempt to determine the direct effects of alcohol on the hypothalamic-pituitary aspect of the hypothalamic-pituitary-gonadal axis, we examined the effects of alcohol on serum luteinizing hormone (LH) levels in the normal and testosterone-depleted castrated male rat. We found that within several days after castration (2-4 days) alcohol, at low to moderate doses, was only modestly effective in suppressing serum LH levels, whereas in sham-operated controls it was maximally effective at all doses tested. Surprisingly, at very high doses (4-6 g/kg), alcohol not only did not depress serum LH levels in long-term castrated rats, but elevated them by 2- to 4-fold when compared to saline-injected controls. The loss of sensitivity of the hypothalamic-pituitary axis to alcohol observed in castrated rats appeared to be selective, inasmuch as the mortality rate at high doses of alcohol was significantly (P less than .01) greater in castrated rats, when compared to sham-operated controls, and other measures of central nervous system impairment were equivalent in both groups. At the present time, it is difficult to explain this biphasic effect of alcohol on serum LH in the castrated animal, but our results are consistent with the hypothesis that the effects of alcohol on gonadotropin release may be dependent to a significant degree on the steroid milieu at the time of the experiments.