Introduction: Stroke including ESUS is one of the most common causes of disability and death in the world. Atrial fibrillation (AF) of more than 30 sec is considered as reason for cardioembolic stroke. Hypothesis: Nevertheless, shorter periods of AF might also be a possible risk factor. Methods: We determined periods of AF of 0-14 sec, AF 15-30sec and AF> 30 sec in patients with different stroke entities. We analysed all ischemic strokes and transient ischemic attacks at the Stroke Unit in Siegen over 6 months. All stroke patients were prospectively ECG monitored over a minimum of 24h and characterized for the different AF periods. Results: A number of 714 patients over 6 months were hospitalized suffering a stroke. Among them 163 (30.8%) had a cryptogenic Stroke, including 98 (19%) ESUS patients. Furthermore, 185 (26%) TIA, 209 (39%) cardioembolic, 110 (21%) atherosclerotic, 40 (8%) lacunar and 7 (1%) other specific strokes were registered. Manifest atrial fibrillation (>30s) showed a prevalence within our stroke population of 23%. Whereas 15% had an AF episode of 15-29 s and 16% presented AF episode of 0-14 s. Among cardioembolic infarcts, 45% had manifest atrial fibrillation (>30 s), 20% had an atrial fibrillation episode of 15-29 s, and 22% showed an atrial fibrillation episode of 0-14 s. Taken together, more than 90% of cardioembolic infarcts show an episode of atrial fibrillation of any duration. An AF episode of 15-29 s or 0-14 s is also found in almost 35% of ESUS patients;similar to cryptogenic infarcts. In 22% of TIA patients manifest atrial fibrillation has been detected. 7% of TIA patients showed one or more episodes of AF of 15-29 s and 10% an episode of 0-14 s. Among the atherosclerotic infarcts, the proportion of AF >30 s is significantly lower at 12%, however the percentage of AF for 15-29 s was 14. Conclusions: In our study, AF showed the largest ratio with 48% among cardioembolic infarcts. Since a manifest AF is an exclusion criterion for ESUS, none suffered from manifest AF. However, it is striking that 15% had an AF episode of 15-29 s and 18% had an AF episode of 0-15 s. Possibly, the definition of AF has to be adjusted or at least advanced rhythm monitoring should be provided. This should be investigated in future prospective studies.