Although varicella vaccine is recommended for infants, many physicians and parents have withheld vaccination from infants because of concerns about the vaccine's long-term efficacy. We used a decision-analytic Markov model to examine the effects of decreasing vaccine efficacy on individuals and society. The model incorporated published data on age-specific incidence, morbidity, and mortality rates, as well as data on shifting disease burden from childhood to adulthood as vaccine compliance increases. The effects of 2 vaccination strategies---vaccinating infants at age 12 months and waiting to vaccinate until children are 10 years of age---were compared with the effects of no vaccination. If the efficacy of the vaccine were to decrease by 75%, then 50% compliance with vaccination at age 12 months would save 1800 life-years and 12,800 quality-adjusted life-years annually in the United States. The quality-adjusted life expectancy of individuals vaccinated at age 12 months would be 63 h longer than that of nonvaccinated individuals and would increase to 79 h as vaccination compliance increases and the burden of chickenpox shifts to adulthood. Varicella vaccination of infants at age 12 months appears to be beneficial, even if the efficacy of the vaccine declines substantially.