This paper uses mathematical models and data analysis to examine the epidemiological implications of possible immunologically mediated links between patterns of varicella and herpes-zoster incidence in human communities. A review of previously published reports does not clarify whether or not there is a relationship between the incidence of varicella and the incidence of zoster. However, new analysis of data collected by the Royal College of General Practitioners provides indirect evidence for the hypothesis that a high intensity of varicella transmission suppresses viral reactivation. The significance of this finding for proposed varicella vaccination campaigns is explored by a review of published data on the use of the vaccine. No significant difference is shown to exist between the risk of zoster caused by the vaccine and the wild virus. A mathematical model is then developed to take into consideration the influence of the prevalence of varicella on viral reactivation and the impact of vaccination with attenuated virus, which may be able to recrudesce. Under some conditions, mass application of such vaccines may have the impact of increasing zoster incidence. The results presented here indicate that, before starting any vaccination programme against varicella, its consequences need to be assessed in much more depth.