Abstract Background Nurse-led integrated care is expected to improve outcome of patients with atrial fibrillation compared with usual-care provided by a medical specialist. Methods and results We randomized 1375 patients with atrial fibrillation (64 ± 10 years, 44% women, 57% had CHA2DS2-VASc ≥ 2) to receive nurse-led care or usual-care. Nurse-led care was provided by specialized nurses using a decision-support tool, in consultation with the cardiologist. The primary endpoint was a composite of cardiovascular death and cardiovascular hospital admissions. Of 671 nurse-led care patients, 543 (81%) received anticoagulation in full accordance with the guidelines against 559 of 683 (82%) usual-care patients. The cumulative adherence to guidelines-based recommendations was 61% under nurse-led care and 26% under usual-care. Over 37 months of follow-up, the primary endpoint occurred in 164 of 671 patients (9.7% per year) under nurse-led care and in 192 of 683 patients (11.6% per year) under usual-care [hazard ratio (HR) 0.85, 95% confidence interval (CI) 0.69 to 1.04, P = 0.12]. There were 124 vs. 161 hospitalizations for arrhythmia events (7.0% and 9.4% per year), and 14 vs. 22 for heart failure (0.7% and 1.1% per year), respectively. Results were not consistent in a pre-specified subgroup analysis by centre experience, with a HR of 0.52 (95% CI 0.37–to 0.71) in four experienced centres and of 1.24 (95% CI 0.94–1.63) in four less experienced centres (P for interaction <0.001). Conclusion Our trial failed to show that nurse-led care was superior to usual-care. The data suggest that nurse-led care by an experienced team could be clinically beneficial (ClinicalTrials.gov NCT01740037). Trial Registration number ClinicalTrials.gov (NCT01740037).