The relationship between premenstrual dysphoria and alcohol acquistion and use patterns was studied in 14 women over 35 days on a clinical research ward. A 7-day drug-free baseline was followed by 21 days of alcohol availability and a post-alcohol drug-free period of 7 days. Women could earn alcohol or 50 cents for 30 min of performance on a simple operant task. The Premenstrual Assessment Form (PAF) was administered during baseline and at the premenstruum. Menstrual cycle phase was associated with changes in alcohol use by 8 of the 14 women. Five women drank more during the premenstruum and scored higher on 12 of 18 PAF factors than women whose drinking remained the same (N = 6) or decreased (N = 3). Women who increased drinking had significantly higher PAF scores on impaired social functioning, hostility/anger and hysteroid features (P less than 0.05, 0.01). Women who drank less during the premenstruum reported more pain and physical discomfort, whereas women who drank about the same amount were relatively asymptomatic. PAF profiles reported at the premenstruum and baseline PAF profiles based on recall of the preceding three menstrual cycles were concordant in 13 of the 14 women (P less than 0.001-0.0001). These data suggest that women's perceptions of premenstrual symptomatology are stable through time.