To examine the effects of a thiazolidinedione (600 mg troglitazone) insulin-sensitizing treatment on total body fat measured by underwater weighing, on intra- and extra-abdominal fat mass using magnetic resonance imaging (MRI), and on anthropometric measures. Type 2 diabetic outpatients were studied in a double-blind randomized trial carried out at Glasgow Royal Infirmary, Scotland. Groups who received troglitazone (8 men, 3 women) and placebo (8 men, 2 women) were well matched for age, BMI, total body fat percentage by underwater weighing, and intra-abdominal fat (kilograms) by MRI. After 12 weeks, body weight changes in the troglitazone group (mean +0.66 kg [95% CI -0.71 to 2.04], P = 0.31) and the placebo group (mean +0.25 kg [-0.64 to 1.13], P = 0.55) were not statistically different. Changes in total body fat with troglitazone (mean +1.02% body wt [-1.13 to 3.17], P = 0.32) and placebo (mean -0.54% body wt [-1.68 to 0.59], P = 0.31) were not significantly different. There was, however, a decrease in intra-abdominal fat mass in the troglitazone-treated group (mean -0.47 kg [-0.79 to -0.13], P = 0.01), and this was significantly different (P = 0.03) from placebo treatment (mean -0.41 kg [-0.77 to -0.05]). Treatment with the thiazolidinedione troglitazone in human patients with type 2 diabetes decreases intra-abdominal fat mass but does not affect total body fat or weight. This potentially valuable effect points to a differential action on insulin sensitivity in different adipose tissue depots.