Body composition was determined by dual energy X-ray absorptiometry (DXA) scanning and classical reference methods (40K, 3H2O, and a combination of these in a four-compartment model) in 19 overweight patients with rheumatoid arthritis who underwent a 12-week weight-reducing regimen. The aim of the study was to investigate whether DXA provides a valid estimate of body composition. The results showed that weight as determined by DXA was highly significantly correlated to weight determined by scales. Furthermore, significant correlations were found in the body components (fat-free mass, fat mass) determined by DXA, 40K, 3H2O and the four-compartment model. Mean values differed slightly but significantly. With respect to changes in body composition, no significant correlations were found between any of the methods, except for the weight loss recorded by DXA and scales, and loss of fat mass (and fat free mass) estimated by 3H2O and the four-compartment model. The sparseness of correlations reflected the small changes in fat-free mass and fat mass (2.6 and 1.7 kg respectively), and the fact that changes were comparable to measurement errors of the various methods. We suggest that DXA scanning is a valid supplement for determination of body composition. Validation of DXA scanning requires new experimental in vitro investigations, which, incidentally, also applies to the classical reference methods.