Fifty immunocompromised children with varicella who exhibited no signs of dissemination were treated with intravenous acyclovir or placebo in a double-blind, randomized study. Twelve of 25 placebo recipients were withdrawn from treatment because of their deteriorating condition and were given open acyclovir therapy; only one of 25 recipients of acyclovir was similarly withdrawn (P less than .001). Among those patients who did not receive open treatment, acyclovir significantly reduced time to full crusting (P = .01). Overall, acyclovir, as judged by the physician, significantly improved the patients' condition.