Seventy-nine children with haemolytic uraemic syndrome (mean age 28 months) were randomly assigned either to a group receiving plasma infusions (plasma group, n = 39) or to a group treated conservatively (control group, n = 40). The duration of haemolysis, thrombocytopenia and anuria was similar in the two groups. Serum creatinine levels were similar in the two groups at the 1-month follow-up but were higher in the control group at 3 months (plasma group 49 +/- 14, control group 66 +/- 28 mumol/l; P less than 0.02) and at 6 months (plasma group 48 +/- 13, control group 63 +/- 21 mumol/l; P less than 0.005). The prevalence of proteinuria was also higher in the control group at the 6-month follow-up (plasma group 17%, control group 46%; P less than 0.02). However, differences were no longer significant after 1 year. Renal tissue was examined in 54 cases (plasma group, n = 27; control group, n = 27). Diffuse cortical necrosis was present in 7 cases in the control group but was absent in the plasma group (P less than 0.02). Taking into consideration the higher serum creatinine levels, the higher prevalence of proteinuria during the first 6 months of follow-up in the control group and the presence of diffuse cortical necrosis in this group compared with the plasma group, we conclude that plasma infusions should be regarded as beneficial. Further study is needed to determine which plasma fraction is involved.