This paper has two main goals. The first is to review the context for studying infant mortality, which includes a review of the theoretical framework, the covariates used to examine mortality over the first 60 months of life, and the major findings of empirical studies. Second, the paper adds some new empirical evidence that comes from the longitudinal reconstitution of church registers of Bejsce parish, located in the south of Poland. This rich database allows for an analysis of mortality trends of cohorts born between the 18th and 20th centuries in the parish. The analysis includes a reconstruction of descriptive measures of infant and childhood mortality, and a hazard model of mortality over the first 60 months of life. The hazard model has been calculated for each cohort separately in order to demonstrate the change in the relative importance of analyzed factors during the process of mortality decline in the parish. Obtained mortality patterns are discussed with reference to the theoretical context presented in the first part of the paper.