Various risk scores have been developed for the assessment of operative risk in cardiac surgery. Although risk stratification has been acknowledged as a useful tool to analyze trends in therapy and changes in patient populations, its relevance in assessing the indication for surgery has been questioned. It was the goal of this prospective study to compare 6 common risk scores with regard to the predictive value for mortality in individual patients. Between September 1998 and February 1999 all adult patients undergoing heart surgery were prospectively scored according to the following scores: initial Parsonnet, Cleveland Clinic, French, Euro, Pons, and the Ontario Province Risk score. Early lethality was assessed within 30 days postoperatively. Follow-up was completed in 504 patients. With the exception of the Ontario Province Risk score, lethality in the high risk group was overestimated by all scores, whereas lethality in low to moderate risk groups was underestimated by several scores. Mean scores of surviving and deceased patients showed a broad overlap with high standard deviations. Preoperative risk scores are effective tools for stratification of patient populations and the analysis of surgical outcome. With the aid of risk scores, operative risk can be sufficiently predicted for patient populations or subpopulations. The Euro score best predicted the outcome of our patients. However, when the indication for surgery is to be determined in an individual patient, risk scores should be only considered as an orientation in the decision process.