For a plausible range of values for the different efficacy characteristics of the live varicella (Oka) vaccine at different levels of coverage, modeling results suggest that routine immunization of preschool children would greatly reduce the number of primary varicella cases, whereas the shift in age distribution of cases would not result in increased overall morbidity as measured by number of hospitalizations. Although information about some of the vaccine assumptions is scanty, the combinations of assumptions leading to an increase in morbidity seem unlikely. A catch-up program in older children who have not yet had chickenpox will be important. The number and age distribution of the cases in vaccinees are sensitive to assumptions about the vaccine, especially the degree and distribution of partial protection against infection, relative residual infectiousness, and waning of immunity. Responsiveness to boosting by wild-type VZV infection was especially important in reducing the number of older cases. The advantage conferred by responsiveness to boosting depends on the level of transmission. The direct and indirect effects of vaccines and vaccination programs interact. Understanding how a vaccine works at the individual level is important for the vaccinated individual, but it also influences the overall public health benefits of an immunization programs. Lieu et al based a cost-effectiveness analysis of varicella vaccines on this model of varicella dynamics and assumptions about how the vaccines work. Models cannot replace biologic understanding. The purpose of such models is not to predict the number of cases of varicella, but to examine some possible consequences of introducing a vaccine into the routine immunization schedule of preschool children in the United States, effects of different vaccination strategies, and the benefits of a temporary catch-up program for older children. Modeling exercises of this sort force us to formalize our thinking, for instance about the vaccine mechanisms, and to admit our uncertainties, such as about the vaccine efficacy assumptions. Such models also show where more data need to be collected, for example, on boosting and waning of immunity and relative residual infectiousness. Improvements in the design of vaccine efficacy studies are necessary to provide the input to these models for looking at the long-term effects of vaccination programs. Frailty models can be used to analyze the data in the presence of heterogeneities in susceptibility. Waning can also be estimated using appropriate methods. Relative infectiousness of vaccinees with breakthrough cases can be measured by comparing the relative secondary attack rates when the index infected person is vaccinated and when the index infected person is unvaccinated. More studies are needed to understand how to evaluate responsiveness to boosting. Vaccine efficacy studies in the field should be designed to obtain better estimates of residual susceptibility, residual infectiousness, duration of protection, and the effects of boosting by reinfection with wild-type VZV.