Background There is increasing debate whether atrial fibrillation (AF) episodes during follow-up in patients with embolic stroke of undetermined source (ESUS) are causally associated with the event. AF-related strokes are more severe than strokes of other etiologies. In this context, we aimed to compare stroke severity between ESUS patients diagnosed with AF during follow-up and those who were not. We hypothesized that, if AF episodes detected during follow-up are indeed causally associated with the index event, stroke severity in the AF group should be higher than the non-AF group. Methods Dataset was derived from the Athens Stroke Registry. ESUS was defined by the Cryptogenic Stroke/ESUS International-Working-Group criteria. Stroke severity was assessed by the National Institutes of Health Stroke Scale (NIHSS) score. Cumulative probabilities of recurrent stroke or peripheral embolism in the AF and non-AF ESUS groups were estimated by Kaplan–Meier analyses. Results Among 275 ESUS patients, AF was detected during follow-up in 80 (29.1%), either during repeated electrocardiogram monitoring (18.2%) or during hospitalization for stroke recurrence (10.9%). NIHSS score was similar between the two groups (5 2-13 versus 5 2-14, P  = .998). More recurrent strokes or peripheral embolisms occurred in the AF group compared with the non-AF group (42.5% versus 13.3%, P  = .001). Conclusions Stroke severity is similar between ESUS patients who were diagnosed with AF during follow-up and those who were not. Given that AF-related strokes are more severe than strokes of other etiologies, this finding challenges the assumption that the association between ESUS and AF detected during follow-up is as frequently causal as regarded.