Background: The aim of this study was to compare the prevalence of metabolic syndrome in human immunodeficiency virus (HIV)-infected patients treated with highly active antiretroviral therapy (HAART), using the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III), European Group for the Study of Insulin Resistance (EGIR), and International Diabetes Federation (IDF) definitions. Methods: A cross-sectional study was carried out with 159 consecutive adult HIV-infected subjects (120 males and 39 females) under HAART. Anthropometric and laboratory parameters were measured by standard methods. Hyperinsulinemia was defined by a fasting concentration >75th percentile of values obtained in healthy individuals (107.5pmol/L). Results: The prevalence of ATP III-defined metabolic syndrome was 10.1%; it was 28.3% according to EGIR criteria and 15.1% using the IDF definition. The concordance between the definitions was low (kappa coefficient ranging between 0.134 and 0.296). All subjects with EGIR-defined metabolic syndrome had hyperinsulinemia, but only 50% of those with ATP III-defined metabolic syndrome and 62.5% in the IDF metabolic syndrome population had hyperinsulinemia. Conclusions: The inclusion of hyperinsulinemia as a criterion in the EGIR metabolic syndrome definition made it more discriminative than the ATP III definition, both in men and women, and than the IDF definition in men to identify metabolic syndrome in HIV-infected subjects under HAART. © Copyright 2011, Mary Ann Liebert, Inc.