OBJECTIVE: We investigated the association of the metabolic syndrome with new-onset diabetes in the Hong Kong Cardiovascular Risk Factor Prevalence Study cohort. RESEARCH DESIGN AND METHODS: We followed up on 1,679 subjects without diabetes at baseline. Those with a previous diagnosis of diabetes or those who were receiving drug treatment were considered to be diabetic. The remaining subjects underwent a 75-g oral glucose tolerance test (OGTT). Diabetes was defined by plasma glucose > or =7.0 mmol/l with fasting and/or > or =11.1 mmol/l at 2 h. RESULTS: The prevalences of the metabolic syndrome at baseline were 14.5 and 11.4%, respectively, according to U.S. National Cholesterol Education Program (NCEP) and International Diabetes Federation (IDF) criteria. After a median of 6.4 years, there were 66 and 54 new cases of diabetes in men and women, respectively. The metabolic syndrome at baseline predicted incident diabetes. Hazard ratios (HRs) for the NCEP and IDF definitions of the syndrome were 4.1 [95% CI 2.8-6.0] and 3.5 [2.3-5.2], respectively. HRs for fasting plasma glucose (FPG) > or =6.1 or 5.6 mmol/l were 6.9 [4.1-11.5] and 4.1 [2.8-6.0], respectively. The NCEP and IDF criteria had 41.9 and 31.7% sensitivity and 87.5 and 90.2% specificity, respectively. Their positive predictive values were low, approximately 20%, but their negative predictive values were approximately 95%. CONCLUSIONS: The metabolic syndrome, particularly its component, elevated FPG, predicts diabetes in Chinese. An individual without the metabolic syndrome is unlikely to develop diabetes, but one who has it should practice therapeutic lifestyle changes and have periodic FPG measurements to detect new-onset diabetes.