It has been suggested that the current dietary recommendations (low-fat–high-carbohydrate diet) may promote the intake of sugar and highly refined starches which could have adverse effects on the metabolic risk profile. We have investigated the short-term (6-d) nutritional and metabolic effects of an ad libitum low-glycaemic index–low-fat–high-protein diet (prepared according to the Montignac method) compared with the American Heart Association (AHA) phase I diet consumed ad libitum as well as with a pair-fed session consisting of the same daily energy intake as the former but with the same macronutrient composition as the AHA phase I diet. Twelve overweight men (BMI 33·0 (SD 3·5) KG/M2) WITHOUT OTHER DISEASES WERE INVOLVED IN THREE EXPERIMENTAL CONDITIONS WITH A MINIMAL WASHOUT PERIOD OF 2 WEEKS SEPARATING EACH INTERVENTION. BY PROTOCOL DESIGN, THE FIRST TWO CONDITIONS WERE ADMINISTERED RANDOMLY WHEREAS THE PAIR-FED SESSION HAD TO BE ADMINISTERED LAST. DURING THE AD LIBITUM VERSION OF THE AHA DIET, SUBJECTS CONSUMED 11695·0 (sd 1163·0) kJ/d and this diet induced a 28 % increase in plasma triacylglycerol levels (1·77 (sd 0·79) v. 2·27 (sd 0·92) mmol/l, P<0·05) and a 10 % reduction in plasma HDL-cholesterol concentrations (0·92 (sd 0·16) v. 0·83 (sd 0·09) mmol/l, P<0·01) which contributed to a significant increase in cholesterol:HDL-cholesterol ratio (P<0·05), this lipid index being commonly used to assess the risk of coronary heart disease. In contrast, the low-glycaemic index–low-fat–high-protein diet consumed ad libitum resulted in a spontaneous 25 % decrease (P<0·001) in total energy intake which averaged 8815·0 (sd 738·0) kJ/d. As opposed to the AHA diet, the low-glycaemic index–low-fat–high-protein diet produced a substantial decrease (-35 %) in plasma triacylglycerol levels (2·00 (sd 0·83) v. 1·31 (sd 0·38) mmol/l, P<0·0005), a significant increase (+1·6 %) in LDL peak particle diameter (251 (sd 5) v. 255 (sd 5) Å, P<0·02) and marked decreases in plasma insulin levels measured either in the fasting state, over daytime and following a 75 g oral glucose load. During the pair-fed session, in which subjects were exposed to a diet with the same macronutrient composition as the AHA diet but restricted to the same energy intake as during the low-glycaemic index–low-fat–high-protein diet, there was a trend for a decrease in plasma HDL-cholesterol levels which contributed to the significant increase in cholesterol:HDL-cholesterol ratio noted with this condition. Furthermore, a marked increase in hunger (P<0·0002) and a significant decrease in satiety (P<0·007) were also noted with this energy-restricted diet. Finally, favourable changes in the metabolic risk profile noted with the ad libitum consumption of the low-glycaemic index–low-fat–high-protein diet (decreases in triacyglycerols, lack of increase in cholesterol:HDL-cholesterol ratio, increase in LDL particle size) were significantly different from the response of these variables to the AHA phase I diet. Thus, a low-glycaemic index–low-fat–high-protein content diet may have unique beneficial effects compared with the conventional AHA diet for the treatment of the atherogenic metabolic risk profile of abdominally obese patients. However, the present study was a short-term intervention and additional trials are clearly needed to document the long-term efficacy of this dietary approach with regard to compliance and effects on the metabolic risk profile.