1. Weight loss, resting metabolic rate and nitrogen loss were measured in forty obese inpatients on reducing diets. 2. Five subjects ate 3.55 MJ/d for 6 weeks (Expt 1). Twenty-one subjects ate 4.2 MJ/d for the first week, 2.0 MJ/d for the second week and 4.2 MJ/d for the third week (Expt 2). Fourteen subjects ate 3.4 MJ/d for the first week and then 0.87 MJ protein or carbohydrate for the second or third weeks, using a cross-over design for alternate patients (Expt 3). 3. Patients in Expt 1 had highest weight loss and N loss in the first 2 weeks, but adapated to the energy restriction over the remaining weeks. On average subjects were in N balance at the end of the study. 4. In Expt 2 patients eating 2.0 MJ/d in week 2 showed increased weight loss compared with week 1. N loss was not raised but it failed to decrease as it had in Expt 1. Weight loss and N loss were reduced on return to 4.2 MJ/d for a third week. 5. In Expt 3 patients eating 0.87 MJ protein showed significantly more weight loss and less N loss than patients eating 0.87 MJ carbohydrate. 6. Resting metabolic rate decreased with time on the low-energy diet, but the manipulations of energy or protein content did not significantly affect the pattern of decrease. 7. Both weight loss and N loss were greater the lower the energy intake, and both decreased with time. Diets with a high protein:energy value give a favourable value for N:weight loss at each level of energy intake.