Atrial fibrillation (AF) is a common and treatable cause of ischemic stroke, but it can be paroxysmal and asymptomatic, and therefore difficult to detect. Patients with stroke routinely undergo 24 hours of continuous cardiac telemetry during hospitalization for stroke as a means of excluding AF. Small studies indicate that extending the duration of monitoring with portable outpatient telemetry devices detects more cases of AF. However, these studies are small and lack control groups, and cannot demonstrate that prolonged cardiac monitoring detects more cases of AF than routine clinical follow-up. The investigators therefore propose a pilot study to determine the feasibility of randomizing patients to prolonged cardiac monitoring or routine clinical follow-up. The investigators will enroll 40 consecutive adult patients seen at the University of California at San Francisco (UCSF) Neurovascular service with cryptogenic stroke or high-risk TIA (ABCD2 score 4 or greater). Enrolled patients will be randomized in a 1:1 fashion. Group A will be assigned to wear an ambulatory cardiac event monitor for 21 days. Group B will be discharged home without a monitor and will serve as controls during routine clinical follow-up. The investigators' primary outcome will be feasibility, defined as more than 80% of randomized patients completing full clinical follow-up and more than 70% of cardiac monitoring if applicable. The investigators' secondary outcomes will be diagnoses of AF at 90 days and 1 year and diagnoses of recurrent stroke at 1 year.