Longitudinal effects of a very low-carbohydrate (VLC) and a calorie-matched high-carbohydrate (HC) weight reduction diet were compared in dietary obese Sprague-Dawley rats exhibiting impaired glucose tolerance and insulin resistance. Obese rats were divided into weight-matched groups: (i) VLC rats consumed an energy-restricted 5% carbohydrate, 60% fat diet for 8 weeks, (ii) HC rats consumed an isocaloric 60% carbohydrate, 15% fat diet, and (iii) HF rats consumed a high-fat diet ad libitum. HC and VLC rats showed similar reductions in body fat and hepatic lipid at the midpoint of the weight-reduction program, indicating effects due to energy deficit. At the end point, however, HC rats showed greater reductions in total and percent body fat, hepatic lipid and intramuscular lipid than did VLC rats, suggesting that diet composition induced changes in the relative efficiencies of the HC and VLC diets over time. HC rats showed marked improvement in glucose tolerance at the midpoint and end point, whereas VLC rats showed no improvement. Impaired glucose tolerance in VLC rats at the end point was due to insulin resistance and an attenuated insulin secretory response. Glucose tolerance in energy-restricted rats correlated negatively with hepatic and intramuscular lipid levels, but not visceral or total fat mass. These findings demonstrate that adaptations to diet composition eventually enabled HC rats to lose more body fat than VLC rats even though energy intakes were equal, and suggest that the elevated levels of hepatic and intramuscular lipid associated with VLC diets might predispose to insulin resistance and impaired glucose tolerance despite weight loss.