Background: There is no research on the association of television (TV) watching with atrial fibrillation (AF). Methods: From 1987 to 1989, the authors obtained information on the frequency of TV watching in 14,458 participants, aged 45–64 years, without a history of AF. The authors used the Cox proportional hazards model to estimate hazard ratios and their 95% confidence intervals of AF according to the frequency of TV watching (“never or seldom,” “sometimes,” “often,” or “very often”). Results: During the 294,553 person-years of follow-up, the authors identified 2,476 AF events. Adjustment for other potential confounding factors, including physical activity, did not change the associations, in which “very often” watching TV carried 1.28 (95% confidence interval, 1.09–1.50) times AF risk compared with “never or seldom” watching TV (P for trend = .002). Even among individuals who met a recommended level of physical activity, watching TV “very often” carried 1.36 (1.02–1.82) times AF risk, compared with watching TV “never or seldom.” Conclusion: Greater frequency of TV watching was independently associated with increased risk of AF even after adjusting for physical activity. Moreover, a recommended level of physical activity did not eliminate the increased risk of frequent TV watching for AF. Avoiding frequent TV watching might be beneficial for AF prevention.