Warfarin is underused for stroke prevention in atrial fibrillation(AF). Previous studies addressing this have lacked longitudinal assessment. This study sought to characterize contemporary warfarin use in new- onset AF and evaluate its change over time. It was hypothesized that AF recurrence has an important influence on warfarin use patterns. One thousand five adults from 17 centers in the United States and Canada were enrolled into a prospective observational registry after their first documented episodes of AF. Detailed demographic, clinical history, and management data were collected on all subjects at enrollment, including medication use. Patients were followed at regular intervals for interim events and changes in AF management. Warfarin use at baseline and last follow- up(mean 25± 8 months) after enrollment was modeled using multivariate analysis. Initially, 65% of subjects were prescribed warfarin, but only 44% were taking it at 30 months. Even in “ ideal” candidates for warfarin, the rate of warfarin prescription decreased from 70% at baseline to 50% at 30 months. Stroke risk factors, including hypertension, congestive heart failure, valvular heart disease, and previous stroke or transient ischemic attack were significant predictors of warfarin prescription at baseline. At last follow- up, the relation between AF recurrence and warfarin use(odds ratio 2.3, 95% confidence interval 1.6 to 3.1) was stronger than that for any individual stroke risk factor. In conclusion, predictors of warfarin use in patients with AF include AF recurrence and selected stroke risk factors. The discontinuation of warfarin in a large number of patients with AF over time is a cause for concern in light of data from recent clinical trials. Warfarin is underused for stroke prevention in atrial fibrillation (AF). Previous studies addressing this have lacked longitudinal assessment. This study sought this characterize contemporary warfarin use in new- onset AF and evaluate its change over time. It was hypothesized that AF recurrence has an important influence on warfarin use patterns. One thousand five adults from 17 centers in the United States and Canada were enrolled into a prospective observational registry after their first documented episodes of AF. Detailed demographic, clinical history, and management data were collected on all subjects at Patients were followed at regular intervals for interim events and changes in AF management. Warfarin use at baseline and last follow-up (mean 25 ± 8 months) after enrollment was modeled using multivariate analysis. Initially, 65% of subjects were prescribed warfarin, but only 44% were taking it at 30 months. Even in “ideal” candidates for w arfarin, the rate of warfarin prescription decreased from 70% at baseline to 50% at 30 months. Stroke risk factors, including hypertension, congestive heart failure, valvular heart disease, and previous stroke or transient ischemic attack were significant predictors of warfarin prescription at baseline At last follow-up, the relation between AF recurrence and warfarin use (odds ratio 2.3, 95% confidence interval 1.6 to 3.1) was stronger than that for any individual stroke risk factor. In conclusion, predictors of warfarin use in patients with AF include AF recurrence and selected stroke risk factors. The discontinuation of warfarin in a large number of patients with AF over time is a cause for concern in light of data from recent clinical trials.