Background. The prevalence of metabolic syndrome (MetS) within individual cohorts varies with the definition used. The aim of this study was to compare the prevalence of MetS between IDF and revised NCEP ATP III criteria in an urban Sri Lankan population and to investigate the characteristics of discrepant cases. Methods. 2985 individuals, aged 35-65 years, were recruited to the study. Anthropometric and blood pressure measurements and laboratory investigations were carried out following standard protocols. Results. Age and sex-adjusted prevalences of MetS were 46.1% and 38.9% by revised NCEP and IDF definitions, respectively. IDF criteria failed to identify 21% of men and 7% of women identified by the revised NCEP criteria. The discrepant group had more adverse metabolic profiles despite having a lower waist circumference than those diagnosed by both criteria. Conclusion. MetS is common in this urban Sri Lankan cohort regardless of the definition used. The revised NCEP definition was more appropriate in identifying the metabolically abnormal but nonobese individuals, especially among the males predisposed to type 2 diabetes or cardiovascular disease. Further research is needed to determine the suitability of the currently accepted Asian-specific cut-offs for waist circumference in Sri Lankan adults.