OBJECTIVES: To develop a new, simple, and practical risk index for patients undergoing coronary artery bypass graft (CABG) surgery, to develop a preoperative risk index that is predictive of surgical-site infection (SSI), and to compare the new risk indices with the National Nosocomial Infections Surveillance (NNIS) System risk index. DESIGN: Potential risk factor and infection data were collected prospectively and analyzed by multivariate analysis. Two new risk indices were constructed and then compared with the NNIS System risk index for predictive power for SSI. SETTING: Alfred Hospital is a 350-bed, university-affiliated, tertiary-care referral center. The cardiothoracic unit performs approximately 650 CABG procedures per year. PATIENTS: All patients undergoing CABG surgery within the cardiothoracic unit at Alfred Hospital between December 1, 1996, and September 29, 2000, were included. RESULTS: Potential risk factor data were complete for 2, 345 patients. There were 199 SSIs. Obesity (odds ratio [OR], 1.78; 95% confidence interval [CI95], 1.24 to 2.55), peripheral or cerebrovascular disease (OR, 1.64; CI95, 1.16 to 2.33), insulin-dependent diabetes mellitus (OR, 2.29; CI95, 1.15 to 4.54), and a procedure lasting longer than 5 hours (OR, 1.75; CI95, 1.18 to 2.58) were identified as independent risk factors for SSI. With the use of a different combination of these risk factors, two risk indices were constructed and compared using the Goodman-Kruskal nonparametric correlation coefficient (G). Risk index B had the highest G value (0.3405; CI95, 0.2245 to 0.4565), compared with the NNIS System risk index G value (0.3142; CI95, 0.1462 to 0.4822). The G value for risk index A, constructed from preoperative variables only, was 0.3299 (CI95, 0.2039 to 0.4559). CONCLUSION: Two new risk indices have been developed. Both indices are as predictive as the NNIS System risk index. One of the new risk indices can also be applied preoperatively.