Thirty-four children with juvenile chronic (rheumatoid) arthritis were recruited to a randomized, double-blind study of deflazacort (an oxazolone derivative of prednisone) vs prednisone. All had been receiving glucocorticoid therapy for at least 1 year and required at least 5 mg/d of prednisolone (usually as 10 mg every 2 days). Thirty-one children completed the study. Bone density trends were measured in the spine by dual photon absorptiometry and in the forearm by single photon quantitative computed tomography at 3-monthly intervals. Trends (velocities) in bone and soft tissue growth were calculated. In the spine, bone growth correlated well with indices of soft tissue growth, but covariance analysis showed a significant advantage (P less than .007) of deflazacort when spinal bone mineral growth was compared to body surface area and weight. In part this was due to a temporary interruption in weight by children receiving deflazacort, whose gain in height was comparable with that of the prednisone group. Some children in both groups improved clinically and showed catch-up growth; in these children relative spinal bone mineral growth velocities were about twice those observed for height and weight. It is concluded that during the first year of deflazacort, their spinal bone mineral content at a level that was appropriate for their height and weight. Further observations are required to establish whether this advantage can be maintained subsequently. The anti-inflammatory effects of the two glucocorticoids appeared similar.