We describe the development and application of a modified version of the radiologic scoring method proposed by Larsen in patients with rheumatoid arthritis (RA). We modified Larsen's method adding a semiquantitative description of the loss of joint surface area and provide standardized reference films for all stages at different anatomical sites (metacarpophalangeal joints, proximal interphalangeal joints, wrists, metatarsophalangeal joints). To evaluate the method, standard anteroposterior radiographs of hands, wrists and forefeet of 24 patients with early erosive RA taken at baseline (t0) and after 36 months (t1) were read by 2 raters in a blinded fashion. The interrater difference was compared with the interpatient (t0 to t1) difference using a hierarchical analysis of variance. Moreover, the method was applied to patients included in a 2 year clinical trial to evaluate the efficacy of intramuscular methotrexate (MTX) and gold sodium thiomalate (GSTM) with radiographs taken at baseline and after 6, 12, and 24 months. There was a good agreement between the 2 raters and the change could be well documented. The estimation of the intrapatient variance (at t0 and t1) was 8 times higher than the estimation of the interreader variance. The method was easily applicable in 57 patients treated with MTX and 53 patients under treatment with GSTM, showing a slowing of radiologic progression after Month 6 with both drugs. The modification of Larsen's scoring method is a reliable measure to assess baseline status and radiologic progression in patients with RA.