We evaluate the ability of the metabolic syndrome (MetS) defined by five definitions for predicting both incident CHD and diabetes combined, diabetes alone, and CHD alone in a Chinese population. The screening survey for type 2 diabetes was conducted in 1994. A follow-up study of 541 high-risk non-diabetic individuals who were free of CHD at baseline was carried out in 1999 in Beijing area. The MetS was defined by the World Health Organization (WHO), European Group for the Study of Insulin Resistance (EGIR), American College of Endocrinology (ACE), the International Diabetes Federation (IDF), and the National Cholesterol Education Program and the American Heart Association (AHA) (updated NCEP) criteria. From a multiple logistic regression adjusting for age, sex, education, occupation, smoking, family history of diabetes, and total cholesterol, the relative risk of the ACE-defined MetS for incident diabetes alone (67 cases) was 2.29 (95% CI, 1.20-4.34). The MetS defined by the five definitions was associated with a 1.8-3.9 times increased risk for both incident CHD and diabetes combined (59 cases), and with a 1.9-3.0 times for total incident diabetes (126 cases). None of the five definitions predicted either incident CHD alone (177 cases) or total incident CHD (236 cases). In conclusion, the MetS defined by the current definitions appears to be more effective at predicting incident diabetes.