To obtain a model for the prediction of acute renal failure (ARF) after coronary surgery, 2009 consecutive patients were investigated. ARF was defined as a peak postoperative serum creatinine value exceeding the preoperative value by 50% or more or a need for dialysis. A postoperative increase in serum creatinine of less than 50% was associated with an early mortality (< or = 30 days postop.) of 0.4%. Sixteen per cent of the patients increased their serum creatinine by more than 50% and in this group there was a mortality of 1.3%. Twenty-five patients (1.2%) required postoperative haemodialysis because of ARF and of these 11 (44%) died early, whereas another 7 patients with chronic renal failure, requiring both pre- and postoperative haemodialysis, all survived. Peak postoperative serum creatinine and changes from the preoperative value were analyzed and related to clinical variables. Multivariate analysis indicated that high preoperative serum creatinine, high age and postoperative haemodynamic instability were the most important risk factors for developing renal failure. A logistic model including these risk factors versus the probability of developing ARF is presented.