BACKGROUND: Stroke is a rare but devastating complication after coronary artery bypass grafting (CABG) and its prevention remains elusive. We used a case control design to investigate the extent to which preoperative and perioperative factors were associated with occurrence of stroke in a cohort of consecutive patients undergoing myocardial revascularization. METHODS: From April 1996 to March 2001, data from 4,077 patients undergoing CABG were prospectively entered into a database. The association of preoperative and perioperative factors with stroke was investigated by univariate analyses. Factors observed to be significantly associated with stroke in these analyses were further investigated using multiple logistic regression to estimate the strength of the associations with the occurrence of stroke, after taking account of the other factors. RESULTS: During the study period, 4,077 patients underwent CABG and of these 923 (22.6%) had off-pump surgery. Forty-five patients suffered a perioperative stroke (1.1%). Overall there were 46 in-hospital deaths (1.1%), of whom 6 also suffered a stroke. Brain imaging of the stroke patients showed embolic lesions in 58%, watershed in 28%, and mixed in 14%. Multivariate regression analysis identified several preoperative factors as independent predictors of stroke, ie, age, unstable angina, serum creatinine greater than 150 mcg/ml, previous cerebrovascular accident (CVA), peripheral vascular disease (PVD), and salvage operation. When operative risk factors were added to the adjusted model, off-pump surgery was associated with a substantial, but not significant, protective effect against stroke (odds ratio = 0.56, 95% confidence interval 0.20 to 1.55). Survival for stroke patients was 93% and 78% at 1 and 5 years, respectively. CONCLUSIONS: Overall incidence of stroke is relatively low in our series. Age, unstable angina, previous CVA, PVD, serum creatinine greater than 150 mcg/ml, and salvage operation are independent predictors of stroke. These factors should be taken into account when informing each individual patient on the possible risk of stroke and in the decision-making process on the surgical strategy.