OBJECTIVE It has become very important to avoid homologous blood transfusions in today's cardiac surgery. We performed a retrospective analysis to find out preoperative factors to predict the risk for transfusion of red-cell concentrate in cardiac surgery. METHODS This study included 400 consecutive patients undergoing coronary artery bypass grafting. We also included emergency (4 %) and re-operations (8 %). We tried to find out predictive factors for the need of transfusion of red-cell concentrate on the base of logistic regression coefficient and the odds ratio. We looked at the following factors as predictors of transfusion risk: left ventricular ejection fraction < 0.35, age over 70 years, preoperative hemoglobin < 11 g/dl, insulin-dependent diabetes (IDDM), emergency operation, female sex, impaired renal function (creatinine > 1.6 mg/dl), and re-operation. RESULTS In our group, 132 (33 %) patients received transfusion during hospitalization, while 268 (67 %) did not. On average, 2.2 +/- 0.68 units of red-cell concentrate were transfused per patient. In addition, we found a predictive value for transfusion for the following parameters: age > 70 years, preoperative hemoglobin < 11 g/dl, re-operation and ejection fraction < 0.35. We could not find any significantly increased blood transfusion risk in female cases, insulin dependent diabetes mellitus, or impaired renal function. CONCLUSIONS We could show that there is normally no need for blood transfusion in (2/3) of the patients in cardiac surgery according to this study's results. Furthermore, it was obvious that some patient variables can be used predict the risk for perioperative transfusion. Based on these results, the prophylactic administration of aprotinin or the use of a cell saver could be useful in selected patients.