Objective: To compare the ability of biochemical indices of insulin resistance (IR) with metabolic syndrome (MetS) classifications to predict changes in blood glucose control over a 3-year period in overweight and obese subjects. Design: This was a longitudinal, prospective study, with data collected at baseline, 18 and 36 months. Subjects and methods: A total of 175 overweight (body mass index (BMI)425 kg m À2 ) and obese (BMI430 kg m À2 ) subjects were enrolled in the study. The IR indices assessed included fasting insulin concentration, the insulin/glucose-derived indices, homeostasis assessment model of insulin resistance (HOMA-IR) and quantitative insulin sensitivity check index (QUICKI), the insulin/triglyceride-derived McAuley index, plasma adiponectin concentration and the triglyceride (trig) and high-density lipoprotein (HDL)-cholesterol ratio (trig:HDL). The two MetS classifications were assessed according to the definitions of the National Cholesterol Education Program-Third Adult Treatment Panel (NCEP-ATPIII) and the International Diabetes Federation (IDF). The potential of the IR indices and MetS classifications at baseline to predict the development of impaired fasting glucose (IFG) was examined using receiver-operator characteristic (ROC) curve analysis and analysis of variance. Results: Complete data were collected on 158 subjects. In all, 51 (32%) subjects developed IFG during the study. The analysis of variance showed significant differences between the IFG and normoglycaemic group in the baseline values of the McAuley index, trig:HDL, plasma adiponectin concentration and prevalence of the MetS. The ROC curve analysis confirmed this result and showed that the strongest predictors of IFG were baseline trig:HDL and IDF MetS classification, followed in order by the McAuley index, plasma adiponectin concentration and NCEP-ATPIII MetS classification. In contrast, the baseline values of fasting insulin, HOMA-IR and QUICKI did not predict IFG. Discussion: This study showed that the IR indices, derived, in part, from plasma triglyceride concentration, were sensitive predictors for the development of IFG in normoglycaemic overweight and obese subjects. Indices derived from glucose and insulin did not identify this at-risk group. The study also showed that the presence of MetS and its abnormalities of an increased trig:HDL ratio and low plasma adiponectin concentration were all sensitive predictors of IFG.