Two regimens of steroid treatment for the initial attack of idiopathic nephrotic syndrome (NS) in children were compared in a controlled prospective multi-centre study. Long prednisone therapy consisted of 60 mg/m2 per 24 h for 6 weeks, followed by alternate day 40 mg/m2 per 48 h for 6 weeks. The standard prednisone therapy was 60 mg/m2 per 24 h for 4 weeks, followed by 40 mg/m2 per 48 h for 4 weeks. A total of 71 children with an initial attack of idiopathic NS were allocated at random to the two groups. The cumulative rate of patients with sustained remissions after 2 years was significantly higher after the long course than after the standard treatment (49% vs 19%, P = 0.0079). The mean relapse rate per patient at intervals of 3, 6 and 12 months was lower in the long-course prednisone group than in the standard prednisone group, and the proportion of children with frequent relapses during any subsequent 6 months period was lower in the long-course group than in the standard group (29% vs 57%, P = 0.03). Mild side-effects of corticosteroid therapy were observed more frequently after long-course prednisone treatment. It is concluded that long-course prednisone therapy of the initial attack of steroid responsive NS is preferable to the standard regimen because it reduces the rate of subsequent relapses without increasing the risk for severe steroidal side-effects.