Studies have shown an increased risk of cardiovascular outcomes with the metabolic syndrome, but information on predictive properties of the National Cholesterol Education Program Adult Treatment Panel 3 (NCEP) criteria is sparse. The authors used data from the Hoorn population-based study in the Netherlands including 2,484 participants aged 50-75 years examined in 1989 and followed for cardiovascular morbidity and mortality through 2000 to assess NCEP criteria, excluding known diabetes or cardiovascular disease. Cluster analyses explored whether NCEP identifies a mixture of heterogeneous groups. For each gender, participants meeting NCEP criteria seemed to be divided into clusters distinguished primarily by triglycerides or high density lipoprotein cholesterol. Cutpoints for components predicting cardiovascular events using classification and survival tree methodology varied by endpoint and gender, but Cox model hazards ratios were relatively comparable regardless of cutpoints (range: 1.3-2.5). Clear gradation in risk of cardiovascular outcomes was evident with increasing number of components, with statistically elevated risk for ≥3 (NCEP) components in men but for ≥2 components in women. Exploratory analyses of alternative metabolic syndrome criteria suggest cardiovascular risk estimates comparable to those derived by using NCEP, but criteria evaluating risk on more of a continuum would potentially allow consideration of alternative definitions by gender or for patients with other risk factors. Copyright © 2005 by the Johns Hopkins Bloomberg School of Public Health. All rights reserved.