Purpose This paper aims to compare the prevalence of metabolic syndrome (MetS) on the basis of three criteria. The diagnostic criteria adopted were those of the International Diabetes Federation, the National Cholesterol Education Program – Adult Treatment Panel III and the American Heart Association/National Heart, Lung and Blood Institute.. Design/methodology/approach A transversal study was undertaken with 691 university students in Fortaleza, Brazil, in 2011-2013. Findings The prevalence of MetS varied considerably according to the criteria used, it being 4.1 per cent for the IDF, 0.7 per cent for the NCEP ATPIII and 1.7 per cent for the revised NCEP ATPIII. The criteria of the IDF presented reasonable agreement in relation to the NCEP ATP III (0.294) and revised NCEP ATP III (0.334). Moderate agreement was found between the NCEP ATPIII/revised NCEP ATPIII. Originality/value There is a need for a universal diagnostic criterion for MetS to obtain uniform and more reliable data for the elaboration of public health policies.