In the last few decades several clinical studies evaluated the efficacy and safety of different strategies for antithrombotic prophylaxis to prevent thromboembolic events in patients with atrial fibrillation (AF). Nowadays, a frequently debated point is related to the high embolic risk deriving from the asymptomatic and symptomatic AF recurrence after cardioversion or in paroxysmal AF, especially in patients with a large number of prolonged episodes of AF. In fact, after the recent AFFIRM and RACE trials, patients after successful cardioversion at risk for thromboembolism could also need lifelong anticoagulation. Considering this, should we anticoagulate all patients with clinical risk factors for thromboembolism with a single episode of AF, without considering the hemorrhagic risk? Based on recent trials, it is reasonable to hypothesize that long AF recurrences (> 48 h), both symptomatic and asymptomatic, are present mostly (if not exclusively) in patients with structural left atrial appendage (LAA) dysfunction and remodeling. Conversely, AF recurrences in patients without LAA dysfunction and remodeling, could be too short to allow thrombi formation in the LAA, and the anticoagulation could also be avoided. Once other clinical and echocardiographic determinants of stroke have been excluded, the LAA velocity could select patients with a normal appendage function at low embolic risk who could benefit from anti-aggregation and patients with irreversible appendage dysfunction, at high embolic risk, who need lifelong anticoagulation.