Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Swiss National Science Foundation Introduction Despite the broad use of oral anticoagulants, stroke remains one of the most serious complications in atrial fibrillation (AF) patients. Stroke has been linked to disturbances of the autonomic nervous system as both share similar risk factors. Therefore, impaired cardiac autonomic function may indicate an enhanced stroke risk. Purpose We hypothesized that impaired cardiac autonomic function, quantified by means of heart rate variability (HRV), might be useful in predicting stroke in patients with AF. Methods We enrolled 1,933 patients with a documented history of AF from the multicenter Swiss-AF cohort study who were either in sinus rhythm (SR-group, n=1130) or AF (AF-group, n=803) on a 5-minute resting ECG recording. HRV triangular index (HRVI), standard deviation of normal-to-normal intervals, root mean square root of successive differences of normal-to-normal intervals, mean heart rate, 5-min total power and power in the high frequency, low frequency and very low frequency range were calculated. We constructed cox regression models to analyze the predictive power of HRV parameters for the composite endpoint stroke or systemic embolism. Results Mean age was 71±8 years in the SR group and 75±8 in the AF group, 28% of the total study cohort were female. 36 patients in the SR group (3.2%) and 58 patients in the AF group (6.5%) experienced a stroke or systemic embolism during a follow-up time of 4.0±1.3 years. In patients with sinus rhythm, HRVI <15 was the only HRV parameter independently associated with stroke or systemic embolism (hazard ratio 2.94; 95% confidence interval 1.3-6.8; p=0.011) after adjustment for multiple clinical confounders (age, sex, AF type, history of hypertension, history of diabetes, history of chronic kidney disease, prior stroke or transient ischemic attack, intake of oral anticoagulation, antiarrhythmics or betablockers). In the AF group, no HRV parameter was associated with the composite endpoint. Conclusions HRVI measured during SR on a single 5-minute ECG recording is an independent predictor of stroke or systemic embolism in AF patients. HRV analysis may help to improve risk stratification in AF patients.