BACKGROUND Acute renal failure requiring dialysis (ARF-D) occurs in 1.5% of patients following cardiac surgery, and remains a cause of major morbidity and mortality. While some preoperative risk factors have been characterized, the influence of preoperative and intraoperative factors on the occurrence of ARF following cardiac surgery is less well understood. METHODS Preoperative and intraoperative data on 2843 consecutive adult patients undergoing cardiac surgery with cardiopulmonary bypass (CPB) from February 1, 1995 to February 1, 1997 were recorded and entered into a computerized database. Two definitions of renal failure were employed: (i) ARF defined as a rise in serum creatinine (Cr) of 1 mg/dl above baseline; and (ii) ARF-D defined as the development of ARF for which some form of dialytic therapy was required. The association between preoperative and intraoperative variables and the development of ARF was assessed by multivariate logistic regression. RESULTS A total of 2672 of the 2844 patients underwent isolated coronary artery bypass grafting (CABG) surgery, the remaining 172 underwent valve surgery with or without bypass grafting. Of the CABG patients 7.9% developed ARF and 0.7% developed ARF-D. The mortality for patients who developed ARF was 14% (OR 15, P = 0.0001) compared with 1% among those who did not develop ARF. The mortality for CABG patients who developed ARF-D was 28% (OR 20, P = 0.0001) compared with 1.8% among those who did not require dialysis. Variables that were significantly associated with the development of ARF by multivariate analysis included: increased age, elevated preoperative serum Cr, duration of CPB, presence of a carotid artery bruit, presence of diabetes, reduced cardiac ejection fraction and increased body weight. Variables independently associated with ARF-D included serum Cr, duration of CPB, carotid artery bruit and presence of diabetes. The utility of these models for predicting the development of ARF and ARF-D was confirmed by bootstrapping techniques. Because of the small number of patients who underwent valve surgery, none of these variables were significantly associated with the development of ARF or ARF-D in this group of patients. CONCLUSIONS The development of ARF or ARF-D is associated with a high mortality following CABG surgery. We have identified perioperative variables, which may be useful in stratifying risk for the development of ARF.