The authors retrospectively screened 57,218 consecutive patients for radiologic evidence of stroke within 30 days after noncardiac procedures at a tertiary care university hospital. Incidence of perioperative stroke within 30 days of surgery and associated risk factors were assessed. Patients taking either metoprolol or atenolol were matched based on a number of risk factors for stroke. Parsimonious logistic regression was used to generate a preoperative risk model for perioperative stroke in the unmatched cohort.