Background: Cerebral microbleeds (CMB) incidentally found during an MRI evaluation of ischemic stroke present a therapeutic dilemma. Case control studies suggest a relationship of cerebral microbleeds and intracerebral hemorrhage. Limited prospective studies of patients with ischemic stroke show variability in whether CMB hemorrhage rates are greater than recurrent ischemic stroke rates. Our aim was to determine the prevalence and characteristics of CMB in patients with atrial fibrillation presenting with stroke and secondly what the rate of prospective intracerebral hemorrhage (ICH) was in those same patients. Methods: Consecutive patients admitted with atrial fibrillation and stroke were identified in our institution’s Get with the Guideline database between 2008-2009. Data from the incident hospitalization included: demographic information, use of antithrombotic and/or anticoagulation, and cerebrovascular risk factors. Radiologic imaging was reviewed to determine if a GRE or SWI sequence was included and record the number and location of CMB. Follow up information was obtained by chart review and written survey or phone call. Death certificates were obtained where applicable. Results: 240 patients were identified, 128 of whom had a CT only and 112 who had an MRI. Of those with an MRI, 76 had GRE or SWI sequences. Twenty-one of 76 patients with a GRE or SWI sequence had a CMB (prevalence: 27.6%). One patient in the CMB (+) group had a prospective hemorrhage during an average follow up of 3 years (rate: 0.33 ICH per patient year of follow up). No patients in the CMB negative group had an ICH. Conclusions: Our study demonstrates the prevalence of cerebral microbleeds is much higher than the actual prospective hemorrhage rate. The prospective hemorrhage rate is low and consistent with findings from large, randomized anticoagulation studies in patients with atrial fibrillation.