To determine the utility of steroid response in classifying childhood nephrotic syndrome, we reviewed 119 biopsies in 92 children aged 1 to 16 years who had been followed for a mean of 7.2 years. Steroid responses were classified as steroid resistant, steroid dependent, and frequent relapser as defined by the International Study of Kidney Disease in Children. Biopsy specimens were classified as showing focal glomerulosclerosis (FSGS) in 39 children, as showing lipoid nephrosis in 28, and as questionable in another 25 with either focal global sclerosis, IgM nephropathy, or mesangial prominence and tubular changes. A strong agreement (p less than 0.01) was found between children whose FSGS was steroid resistant and children whose lipoid nephrosis resulted in frequent relapses. The length of the remission after therapy with chlorambucil or cyclophosphamide was determined in 84 children. A significantly shorter length of remission after cytotoxic drug therapy (p less than 0.05) was identified for patients with FSGS versus those with lipoid nephrosis; this difference became more significant for steroid-resistant patients in comparison with those who were steroid dependent or were frequent relapsers (p less than 0.005). Among all steroid-resistant patients, those with FSGS had shorter remissions than patients with other histologic changes (p less than 0.001). The data suggest that patterns of response to corticosteroid therapy correlate with the histologic abnormality. Thus steroid-sensitive patients need not undergo renal biopsy before receiving cytotoxic drugs. Steroid-resistant patients would benefit from a biopsy, because the findings tend to predict the outcome.