To determine the safety and efficacy of varicella vaccine, we studied 437 children in remission from leukemia who were immunized with live attenuated varicella virus. Three hundred one of the patients received two doses of vaccine and 136 received a single dose of vaccine from 1 of 10 lots from two manufacturers. The patients have been followed for an average of three years (range, one to six). Seroconversion occurred in 88 percent of the 437 children after the first dose of vaccine and in 98 percent after one or two doses. The proportions of patients who were seronegative after one, three, and five years were 20, 25, and 30 percent, respectively, with little change over time in the geometric mean titers of specific antibody (6.3, 6.5, and 5.7, respectively). Chickenpox has been documented in 36 vaccinated patients (8 percent) who had 3 to 640 vesicles (mean, 100), mild illness, and no complications. Of the 83 vaccinated patients exposed to varicella within their families, 11 had chickenpox; the attack rate was 14 percent (8 percent among seropositive patients, 29 percent among seronegative patients). There was no relation between the time since vaccination and either the attack rate or the severity of the breakthrough illness. Two doses of vaccine appeared to be no more effective than a single dose. Of the 372 patients receiving maintenance chemotherapy when immunized, 149 (40 percent) had a rash, which was treated with acyclovir in 16 children (4 percent) and became a severe febrile illness in 4. These reactions were not fatal and were all associated with vaccine lots, the use of which has since been discontinued. We conclude that in children in remission from leukemia, varicella vaccine is safe and induces an immunity to chickenpox that persists for more than three years.