To examine the influence of moderate-intensity (50% of VO(2peak)) exercise training (MI) versus high-intensity (75% of VO(2peak)) exercise training (HI) on regional fat distribution and plasma adiponectin, we randomized 18 overweight (body mass index [BMI] = 30 +/- 1 kg/m(2)) elderly (71 +/- 1 years) to HI, MI, or a control group (CON). Subjects enrolled in HI or MI completed a 12-week exercise training protocol designed to expend 1000 kcal/week. Body composition testing was completed prior to and following the exercise training using dual energy X-ray absorptiometry and a computed tomography scan. Plasma adiponectin was measured using enzymelinked immunoassay (ELISA). VO(2peak) improved in HI and MI, whereas there was no change in VO(2peak) in CON. No significant change in body weight, BMI, and % fat occurred in MI, HI, or CON. Although there was a significant reduction in visceral fat with HI (-39 cm(2)), there was no change in the MI or CON groups. In addition, there was a significant increase in thigh muscle attenuation in the HI group. There were no changes in thigh muscle attenuation in the MI and CON groups. Also, there was no change in plasma adiponectin in the MI, HI, or CON groups. In summary, our direct comparison of exercise intensity without weight loss promotes the efficacy of HI in the reduction in visceral fat, even without changes in adiponectin.