Atrial fibrillation (AF) is the most frequently encountered cardiac arrhythmia. It is a risk factor for stroke and premature death. We studied the temporal changes in the prevalence of AF from 1976 to 1994 in a random population aged 50 to 89 years. The prevalence of AF, diagnosed from electrocardiograms (ECGs), was determined in 8,606 patients examined in 1976 to 1978, in 8,943 patients examined in 1981 to 1983, and in 6,733 subjects examined in 1991 to 1994. Changes in prevalence of AF were estimated by logistic regression analysis. In men, the age-standardized prevalence of AF increased from 1.4% in 1976 to 1978 (odds ratio [OR] 1.0, reference) to 1.9% in 1981 to 1983 (OR 1.6, 95% confidence interval [CI] 1.1 to 2.1), and to 3.3% in 1991 to 1994 (OR 2.3, 95% CI 1.6 to 3.4, p<0.001, adjusted for age). In women, the prevalence of AF decreased from 1.5% in 1976 to 1978 (OR 1.0, reference) to 1.0% in 1981 to 1983 (OR 0.7, 95% CI 0.5 to 1.0), and to 1.1% in 1991 to 1994 (OR 0.7, 95% CI 0.5 to 1.0), although the overall decrease was not significant (p=0.11, adjusted for age). After adjusting for changes in comorbidity, body weight, and height, the increase in the prevalence of AF in men from 1976 to 1978 and from 1991 to 1994 remained significant (OR 1.9, 95% CI 1.3 to 2.8, p=0.002). Although unchanged in women, the prevalence of AF in men more than doubled from the 1970s to the 1990s. The factors responsible for this gender-specific increase in the prevalence of this common arrhythmia have yet to be identified.