We retrospectively investigated, in 19 children with systemic-onset juvenile chronic arthritis (SoJCA), the possible influence on the outcome of methotrexate (MTX) therapy of several independent variables, including age at onset of juvenile chronic arthritis, disease duration and severity of the disease at baseline. The dosage of MTX ranged from 7.5 to 11.0 mg/m2/week (median 9.3 mg/m2/week) and was given as a single, oral weekly dose. After 6 months of treatment, 12 (63%) patients were judged as responders on grounds of a > or = 50% reduction in the number of joints with active arthritis and/or an articular severity score; 7 (37%) did not respond to therapy. When the baseline values of the selected variables were compared, we found that the responder group had, with respect to the non-responder group, a lower percentage of radiographic lesions (p < 0.005), a shorter duration of the disease (p < 0.05) and a lower number of joints with limitation of motion (p < 0.01), functional limitation score (p < 0.05) and articular severity score (p < 0.05). A threshold value of disease duration of two years and the presence/absence of radiographic lesions gave a correct classification with respect to the treatment outcome of 73.7% and 83.3%, respectively. The predictive value of these two variables was confirmed by a multivariate analysis. We conclude that earlier treatment with MTX, possibly before the appearance of radiographic changes, may favourably influence the outcome of MTX treatment in those patients with SoJCA who require a second-line drug.