Human leukocyte interferon was evaluated as a treatment for varicella in a randomized double-blind, placebo-controlled study carried out in two phases. A total of 44 children being treated for cancer were enrolled within 72 hours of the appearance of the exanthem. The mean number of days of new lesion formation was 3.8 +/- 1.89 (+/- S.D.) in the interferon recipients and 5.3 +/- 2.56 in the placebo recipients (P less than 0.05). Eighty-one per cent of the interferon recipients had had no new lesions for 24 hours by Day 7, as compared with 56 per cent of the placebo recipients (P less than 0.025). In the second, higher-dose phase of the study 92 per cent of the interferon recipients had had no new lesions for 24 hours by Day 6, as compared with 45 per cent of the placebo recipients (P less than 0.025). Three of 21 placebo recipients died of progressive varicella. Two of the 23 interferon recipients died two to three weeks after the onset of varicella; viral cultures were negative in one of these patients, and the second had recurrent viremia at the end of the treatment period. Among the survivors, treatment with interferon reduced the number of patients who had life-threatening dissemination (none of 21 vs. three of 18; P = 0.053). We conclude that interferon had an antiviral effect against varicella virus in immunocompromised patients.