The aim of this work was to compare and validate seven different methods for estimating changes in fat free mass, in patients suffering from rheumatoid arthritis. Measurements were made of fat and fat free mass before and after 12 weeks on an energy restricted, protein rich diet and physical training. The subjects were sixteen female and three male overweight out-patients (mean body mass index at baseline: 30 kg/m2) suffering from rheumatoid arthritis, according to the criteria of the American Rheumatism Association. Fat free mass was estimated by eight different body composition methods (a four-compartment model, total body water, total body potassium, impedance, near infrared interactance, creatinine excretion, body mass index and skinfold measurements). Mean weight loss was 2.7 kg fat and 1.7 kg fat free mass. There was no difference between measurements of mean change in fat free mass by the four-compartment model and the other methods, except for the creatinine method (P = 0.03). Compared to the four-compartment method, the total body water method gave the most accurate estimate of individual fat free mass changes (residual Mean Square: 0.4 kg), second to this method, the impedance method, seemed most valid (residual Mean Square: 0.8 kg). Accuracies of the other methods were lower (residual Mean Square between 4.2 and 8.2 kg [corrected]). Of eight methods for estimating changes in FFM, the TBW method gave the most accurate estimate of individual FFM changes, compared to a four-compartment model used as reference.(ABSTRACT TRUNCATED AT 250 WORDS)