The present paper analyzes the relation between alcohol intake and serum total estradiol in premenopausal women while attempting to control or reduce several sources of variability of serum estradiol. Sixty premenopausal women were recruited, and alcohol intake was estimated by a semiquantitative questionnaire. Interviews, anthropometric measurements, and blood drawings (after overnight fasting) were conducted twice, 1 year apart. Both blood samples were obtained on the same day of the luteal phase of the cycle, in the same month and in the same hour and minute of the day. Samples from the first drawing were stored at -80 degrees C. Serum from both drawings was assayed simultaneously and in blind fashion. A significant association between alcohol intake and estradiol was found when estradiol was averaged across the two visits (Spearman's r = 0.29; P < 0.05). To control for intraindividual variability of estradiol over time, participants were then divided into tertiles of hormone distribution for each of the two sets of measurements and classified based on their consistency in estradiol across the two visits. Women showing consistently high estradiol levels at both visits were characterized by a significantly higher alcohol intake (92.8 g/week) in comparison with those showing consistently low estradiol at both visits (31.6 g/week). Furthermore, the prevalence of drinkers in the group with consistently high estradiol was significantly higher than in the group with consistently low estradiol. The present report indicates that drinkers seem to be characterized by consistently higher estradiol than nondrinkers, and that when the variability of estradiol in premenopause is considered, it is possible to identify a relationship between alcohol intake and estradiol.