The role of alcohol in recurrences of atrial fibrillation (AF) was assessed in a consecutive series of 98 patients (75 men) aged less than 65 years. In addition to etiologic assessment using clinical and laboratory methods and echocardiography, the patients' drinking habits were evaluated by recording the amount of alcohol used during the week preceding AF, by responses to the CAGE (Cut, Annoying, Guilt, Eye; see below) questionnaire (a screening test for alcohol abuse) and by selected laboratory tests. Two groups of control subjects were studied: 98 sex- and age-matched patients admitted to the emergency ward for acute illnesses, and 50 subjects selected randomly from the local out-of-hospital population. The mean alcohol consumption among men during the study week was 186 g (median 45 g; range 0 to 2,100 g) among patients, whereas among male hospital and population control subjects it was 86 g (30 g; 0 to 1,050 g) and 94 g (35 g; 0 to 630 g), respectively. When the weekly alcohol consumption was analyzed in 3 categories (0; 1 to 210 g; greater than 210 g), there was a significant difference between AF cases and hospital control patients (p = 0.03), but not between AF cases and population control subjects. Multivariate analysis of data of AF cases and population control subjects showed that alcohol intake and a positive response to 1 or more of the CAGE questions were independently related to AF in men. Other independent risk factors were the presence of heart disease, low serum potassium and lack of sleep or experience of excess psychologic stress, or both.(ABSTRACT TRUNCATED AT 250 WORDS)