Aims/hypothesis. Our aim was to test the hypothesis that TNF-α protein levels in skeletal muscle are important in mediating the improvements in glucose homeostasis that are associated with diet and exercise regimens intended to reduce cardiovascular risk. Methods. We recruited 20 people with a body mass index of 32.1±1.2 kg/m 2 (mean ± SEM) and one other component of the metabolic syndrome. The average age was 51.2±8.1 years (mean ± SD). Of the 20 subjects, 6 were men and 14 were women. All subjects completed an 8-week control period, followed by randomisation to 8 weeks of moderate cycling exercise (30 min, three times per week) or to a diet with the following characteristics: low in saturated fat, high in fibre, low glycaemic index, rich in complex carbohydrates.Results. Diet induced a small reduction in body mass index (3.0±0.7%, p<0.05), although weight loss was not intended. Exercise training increased maximum oxygen consumption by 12±6% (p<0.05). Both interventions reduced fasting plasma insulin levels by about 20%. Diet reduced skeletal muscle TNF-α protein by 54±10% (p<0.05), an effect that was independent (p=0.94 in covariate analysis) of the small concurrent weight loss (−2.8±0.7 kg). Levels of GLUT4 protein were unchanged in the diet group. In contrast, exercise training did not significantly change TNF-α protein expression, but GLUT4 protein expression increased by 105±37% (p<0.05). Conclusions/interpretation. These data indicate that the metabolic benefits of a diet aimed at cardiovascular risk reduction are associated with a decrease in skeletal muscle TNF-α protein.