Atrial fibrillation (AF) is the most common sustained dysfunction in heart rhythm clinically and has been identified as an independent risk factor for stroke through formation and embolization of thrombi. AF is associated with reduced cardiac output and short and irregular cardiac cycle length. Although the effect of AF on cardiac hemodynamic parameters has been reported, it remains unclear how the hemodynamic perturbations affect the potential embolization of blood clots to the brain that can cause stroke. To understand stroke propensity in AF, we performed computer simulations to describe trajectories of blood clots subject to the aortic flow conditions that represent normal heart rhythm and AF. Quantitative assessment of stroke propensity by blood clot embolism was carried out for a range of clot properties (e.g., 2–6 mm in diameter and 0–0.8 m/s ejection speed) under normal and AF flow conditions. The simulations demonstrate that the trajectory of clot is significantly affected by clot properties as well as hemodynamic waveforms which lead to significant variations in stroke propensity. The predicted maximum difference in stroke propensity in the left common carotid artery was shown to be about 60% between the normal and AF flow conditions examined. The results suggest that the reduced cardiac output and cycle length induced by AF can significantly increase the incidence of carotid embolism. The present simulations motivate further studies on patient-specific risk assessment of stroke in AF.