Patients with new-onset of atrial fibrillation (NOAF) during acute myocardial infarction (AMI) currently receive long-term oral anticoagulation. The risk for stroke of “early” versus “late” onset of atrial fibrillation (AF) has not been elucidated. Consecutively, AMI patients admitted to a tertiary medical center were analyzed. We excluded patients with preexisting AF, AMI onset ≥24 hours prior to admission, significant valvular disease, fever >38.5°C, in-hospital death, or coronary artery bypass graft. Atrial fibrillation was verified by electrocardiography and medical records. Overall 7061 patients were included, 1.4% developed “early-paroxysmal AF (PAF)” that resolved within 24 hours of admission and 2.5% had “late-AF” beyond the first 24 hours. Median follow-up was ≈6 years. Primary end points included ischemic stroke and all-cause mortality. Stroke rates were higher only in patients with late-AF versus no-AF but not in the early-PAF: 10.6% versus 4.2%, 5.3%, respectively ( P < .001). Death rates were higher in patients with late-AF and early-PAF versus no-AF: 55.3%, 43.2%, and 29.2%, respectively ( P < .001). Congestive heart failure, hypertension, age ≥75, diabetes mellitus, a stroke or transient ischemic attack, vascular disease, age 65-74, female (CHA2DS2-VASc) score underestimated stroke risk in the late-AF group. In conclusion, the study generates the hypothesis that patients with early-PAF may not have a high stroke risk questioning the indication for long-term anticoagulation.