Fat-free mass was measured by hydrodensitometry, electrical impedance and total body potassium before and after water and electrolyte loss induced by (a) the administration of the diuretic frusemide, and (b) sweat loss. All methods of measuring fat-free mass were shown by pilot experiments to have procedural reliability. The diuretic caused a reduction in apparent fat-free mass of 2.63 kg by the impedance method, of 2.33 kg by hydrodensitometry and of 1.8 kg by total body potassium. Water and electrolyte loss from sweating caused a fat-free loss of 2.3 kg, 2.7 kg and 1.3 kg by the same three procedures. Urinary potassium accounted for about one fifth of the observed 40K fat-free mass loss. Each method was thus clearly sensitive to the induced water loss. These data suggest that in evaluating the composition of weight loss, existing methods of measuring body composition do not distinguish between water and other more critical components of fat-free mass. It is thus essential that stable hydration levels are established for any longitudinal comparison of weight loss by these methods.