Childhood disease may have far-reaching effects on adult longevity in high-mortality populations. It was earlier reported from demographic analysis of 19th century New England communities that survivors of high-fatality childhood epidemics lived significantly longer than their unstressed counter-parts. As part of a broader survey of agricultural families of western Massachusetts, these data indicate that elevated child mortality is associated with an increase in the average age at death of survivors. This paper examines this selection process and the problem of heritable longevity using cohort data selected from historical epidemics. A series of tests of the selection hypothesis in which an attempt is made to control demographic covariates suggest that widespread childhood infectious disease strengthens survivors in a developmental manner. Whether such episodes also eliminate less fit individuals according to a simple selection model could not be determined. The influence of family formation variables on child loss, and behavioral (fertility) responses to childhood epidemics in these communities are also considered.