In 1995, varicella vaccination was introduced into the infant immunization schedule of the United States. Currently, many other countries are considering mass varicella vaccination. Mass vaccination has two dangers: it could increase the number of varicella cases in adults, where severity is greater, and increase cases of zoster. A deterministic, realistic, age-structured model (RAS) was built to study these concerns. Model parameter estimates were derived from a review of the literature and surveillance data from England and Wales. Different vaccine efficacy scenarios, vaccine coverages, and vaccination strategies were investigated. The model predicts that, although an upward shift in the age at infection occurs, the overall morbidity due to varicella is likely to decrease following mass infant vaccination. On the other hand, cases of zoster may significantly increase in the first 50 years following vaccination. The model predicts that, in a population similar to England and Wales (50 m people), varicella vaccination with 90% coverage would prevent 0.6 m inpatient days due to varicella but would generate an extra 1.1 m inpatient days due to zoster over the first 65 years. Thus, under base-case model assumptions, the gain in reduction of varicella morbidity from infant vaccination is offset in the short-term by the increases in zoster morbidity (using inpatient days as a proxy). Paradoxically, less effective vaccines or vaccine programmes can be more effective in reducing overall morbidity (varicella + zoster) by allowing the virus to circulate more, which produces a smaller shift in the age at infection and a smaller increase in zoster cases.