To examine whether the live varicella vaccine virus is attenuated, we analyzed varicella vaccine-induced contact cases of clinical chickenpox in healthy siblings of immunized children with leukemia. A rash developed approximately 1 month later in 156 children with leukemia who had been vaccinated. Vaccine-type virus was isolated from 25 of these children. Of 88 known susceptible healthy siblings who were exposed to a vaccine with a rash and from whom follow-up information was available, there was evidence of infection in 15 (17%). Of 15 siblings with seroconversion, 11 (73%) also acquired a mild rash with an average of 38 lesions and no accompanying systemic symptoms. Vaccine-type virus was isolated from four of the contact siblings. Tertiary transmission was documented once. Contact siblings with seroconversion were protected during future household exposure to chickenpox, which occurred in four instances. There was a direct relationship between transmission from vaccinees to varicella-susceptible close contacts and the presence and number of skin lesions in children with leukemia after vaccination. We conclude that in the transmission of varicella, the virus probably originates from skin lesions of infected persons and reaches the respiratory tract of those with secondary cases by the airborne route. On the basis of the mildness of the contact illness, the higher-than-normal rate of subclinical primary infection with varicella-zoster virus in contacts, and the lower-than-normal rate of spread of the vaccine virus to susceptible children in the household, we further conclude that the vaccine virus is attenuated. There was no evidence of reversion of the vaccine virus to virulence.