Atrial fibrillation is the most common arrhythmia and is an independent risk factor for stroke and dementia. Stroke endpoints in atrial fibrillation (AF) studies have been defined differently in the past. Consensus is needed in future AF trials and therefore endpoints should be redefined with accuracy. Depending on type of intervention and patient characteristics in the trial, atrial cardioembolic strokes or all strokes should be used. Atrial cardioembolism is very frequent in AF patients and a combined clinical and imaging approach may help to distinguish cardioembolism from other strokes. Accurate identification of type of stroke is mandatory for the optimal evaluation of interventions in AF patients. To that end, cerebrovascular cardiovascular imaging is instrumental. Intracerebral bleeding is an adverse effect of antithrombotic therapy. Therefore, it is an appropriate primary endpoint in trials evaluating antithrombotic strategies. In studies evaluating combined clinical strategies against stroke, i.e., antiarrhythmic and antithrombotic, as well as strategies focused on maintaining vascular integrity, all major cardiovascular and cerebrovascular events should be included in the endpoint. At present, surrogate endpoints for stroke are not being used. Their role remains to be investigated.