The metabolic syndrome integrates, in a single diagnosis, the manifestations of insulin resistance that may lead to increased cardiovascular morbidity and precedes type 2 diabetes. Here we discuss the strengths and limitations of the definitions of the metabolic syndrome and the epidemiology of the syndrome including information from non-Caucasian populations. The definitions proposed by the World Health Organization (WHO) and the National Cholesterol Education Program (NCEP) are the most frequently used. The relative risk of having long-term complications is greater for the WHO definition; this is explained by the inclusion of the insulin resistance criteria. The cut-off points used in these definitions should be, but are not, adjusted for ethnicity; as a result, in non-Caucasian subjects, there is lack of agreement among these criteria. In a Mexican population-based survey the prevalence was 13.61% using the WHO definition and 26.6% using the NCEP-III criteria. Cases identified by the WHO criteria had a more severe form of the disease. We propose that the metabolic syndrome should be viewed as a progressive long-term process that leads to major complications. Its definition should reflect the continuous nature of the disease; the categorical approach of the current criteria oversimplifies the complexity of the syndrome. The threshold for defining abnormality should be based on the associated risk of the identified phenotype. Refinement of the definition of both affected and nonaffected subjects is required. The available definitions include, in each of these categories, heterogeneous groups with a broad range of risk of future complications.