A multi-centre controlled trial of steroid treatment of the nephrotic syndrome was carried out on 125 patients. Of these, 64 were controls and 61 received prednisone in a recommended dose range of 20-30 mg./24 hours. The actual initial dose averaged 29 mg./24 hours. Treatment was continued for a variable period, but not less than six months. More than 10 mg./24 hours was given on average for 12 months to all patients, and for longer periods to some. Patients were classified, on the basis of biopsy specimens, into three groups: A, minimal change; B, membranous nephropathy; and C, proliferative glomerulonephritis. In groups B and C prednisone did not have any strikingly favourable effect on proteinuria or on renal function as compared with the control group. In group A, however, prednisone reduced proteinuria to a striking and statistically significant extent. It had little if any effect on long-term renal function in any group. The death rate was higher in the combined prednisone groups (17/61) than in the control groups (12/64). This difference was not statistically significant, but there was a significantly higher number of deaths from cardiovascular disease in the prednisone group, whereas the numbers of deaths from renal failure were not significantly different in the two groups.