Background Atrial fibrillation is one of the most frequent complications after cardiovascular surgery. It may result in thromboembolic events, hemodynamic deterioration, and an increased length and cost of hospitalization. Methods We retrospectively studied 504 consecutive adult patients undergoing cardiovascular surgery to determine whether patients with new-onset postoperative atrial fibrillation could be safely discharged in atrial fibrillation after ventricular rate had been controlled and anticoagulation initiated. Results Postoperative atrial fibrillation occurred in 79 (16.2%) of the 487 survivors. Of these patients, 67 were discharged in sinus rhythm, whereas the remaining 12 were discharged in atrial fibrillation. Patients discharged in atrial fibrillation tended to be older, have higher Parsonnet risk scores, and have an increased incidence of valvular heart surgery. Despite this result, this cohort had a shorter length of hospital stay (7.3 ± 2.0 days vs 10.9 ± 9.3 days, p = 0.006), decreased hospital costs ($14,188 ± $2635 vs $23,016 ± $21,963, p = 0.002), and decreased hospital charges ($37,878 ± $7420 vs $58,289 ± $50,980, p = 0.003) compared with patients with atrial fibrillation discharged in sinus rhythm. In the 12 persons discharged home in atrial fibrillation, no repeat hospitalizations, bleeding complications, or thromboembolic events occurred. Conclusion A strategy of early discharge of patients with persistent postoperative atrial fibrillation appears promising and deserves prospective testing on a larger scale. (Am Heart J 1998;135:557-63.)