Reports provide conflicting data about the effects of alcohol consumption on the hormonal system. Any study of these effects must control for a number of variables, including sex, alcohol status (alcoholic addiction vs. non-addiction), medical status (malnutrition, liver disease), and conditions of alcohol exposure, including an acute or continuous pattern of intake. The latter appears to be an especially critical factor in interpreting these effects. The authors therefore conducted a trial with a circadian design in which alcohol was administered repeatedly and regularly over a 26 h period for a total dose of 256 g. Because this protocol involves continuous alcohol administration, it is similar to administration among alcoholics and thus sheds new light on alcohol's effect on hormone secretion. Using healthy volunteers rather than alcoholics, however, prevents any confounding due to liver disorders and nutritional deficiencies, and thus makes it possible to focus on the direct role of alcohol in hormonal modifications. In these conditions, the continuous administration of alcohol did not affect cortisol secretion, but serum testosterone levels were significantly higher at all time points during the alcohol session than at the corresponding time points during the control session. These data are not consistent with previously reported findings for the relation between alcohol and both cortisol and testosterone, because in the current experiment the action of ethanol on steroid secretion should involve the circadian clock more than the hormonal system itself.