Introduction: Cryptogenic stroke comprises about 25% of all ischemic strokes. Depending on modality and duration of ECG monitoring, subclinical atrial fibrillation (AF) is detectable in 2.7-30% of cryptogenic stroke patients. Hypothesis: Extended ECG monitoring after cryptogenic stroke has not been studied in the African American (AA) population. This retrospective study aims to study the incidence and risk factors of subclinical AF in African Americans. Methods: We retrospectively reviewed 96 patients who received implantable loop recorders (ILR) for detecting subclinical atrial fibrillation after cryptogenic stroke. In the vast majority of patients, the ILR was implanted during index hospitalization. Binary univariate and multivariate analyses were performed to determine predictors for AF detection. Results: AF was detected in 29% of patients (28/96) at 1000 days. All AF that was detected was exquisitely paroxysmal and ranged in duration between 0.05-103 minutes (mean 8.4 minutes with SD= 22.1 minutes). Binary univariate analysis revealed the use of non-dihydropyridine calcium-channel blockers to be associated with decreased odds of AF detection. Multivariate analysis found coronary artery disease diagnosis to be associated with increased odds of AF detection. Fifty percent of the events in the AF group were detected within the first 36 days of loop recorder implantation. Conclusions: AF detection in our population occurs very early after index stroke and at significantly higher rates that reported before. Baseline characteristics have a poor predictive ability for the detection of AF. These findings emphasize the need for pre-discharge ILR implantation to improve AF detection in all patients with cryptogenic stroke.