The aim of this study was to investigate the validity of bioelectrical-impedance analysis (BIA) in measuring changes in lean body mass (LBM) that occur in the treatment of obesity with very-low-calorie diets (VLCDs). Resistance (R), reactance (Xc), and nitrogen balance (BN) were measured for 28 consecutive days in 16 obese women consuming isocaloric and isonitrogenous ketogenic and nonketogenic liquid VLCDs while housed in a metabolic ward. Changes in LBM were calculated by using eight BIA-predictive equations and from cumulative BN. Changes in LBM calculated by BN were larger in the ketogenic than in the nonketogenic diet (1.5 +/- 0.1 vs 0.6 +/- 0.2 kg/28 d, P less than 0.05). Neither R nor Xc changed significantly during treatment with either diet but each equation showed losses of LBM with time. All the equations overestimated the losses in LBM calculated by BN. BIA does not appear to be a valid method for measuring the small losses in LBM that occur during treatment of obesity with VLCDs.