By a method using sealed envelopes a series of patients with active chronic hepatitis who had not received corticosteroids or adrenocorticotrophic hormone was divided into two groups; one was given prednisolone and the other was used as a control. Fifty-four patients were initially included in the study over a period of 48 months, but five were subsequently withdrawn when it became clear that the diagnosis was incorrect. The corticosteroid group then contained 22 and the control group 27 patients. The two groups were reasonably well matched; the corticosteroid group, however, contained rather more females of a lower age. Although the mean serum bilirubin was higher in the control group, the mean serum aspartate aminotransferase and total globulin were higher in the corticosteroid group, indicating that that group contained more patients with a high degree of ‘activity’ of the hepatic lesion. The mean erythrocyte sedimentation rate was also higher in the corticosteroid group. An initial dose of 15 mg prednisolone daily was used in the corticosteroid group, and this was reduced whenever biochemical tests of liver function became normal or when there was evidence of severe side-effects of corticosteroids. It was subsequently stopped altogether in eight patients. During the first two to three years after inclusion to the study, the mean serum bilirubin, total globulin and albumin in the corticosteroid group showed a significant trend to normal when compared with the control group. The most significant and unexpected finding was a marked rise in serum albumin. After the investigation had been in progress for 72 months, three patients in the corticosteroid group and 15 in the control group had died at a mean of 11 (4 to 17) and 23 (1 to 63) months respectively. The difference between the mortality in the two groups is significant (P<0·01). The survivors in the two groups had been in the study for 53 (25 to 72) and 52 (29 to 72) months respectively. Four patients in the corticosteroid group had illnesses which were probably severe complications of corticosteroid therapy, and several others had minor complications. It is concluded that corticosteroid therapy should he used during the early-active phase (usually two to three years) of active chronic hepatitis and is of value in increasing life expectancy in patients with the condition. The mode of action is not clear, but it is probably related both to its immuno-suppressive properties and also an ability to stimulate albumin synthesis. The dose should be kept as low as possible, based on serum globulin and albumin concentrations, and be withdrawn when all biochemical tests of liver function return to normal.