Notwithstanding the extensive social scientific research, there are still unanswered questions regarding the persistence of high fertility in sub-Saharan Africa. Although fertility behaviour depends on multiplicity of factors, childhood mortality is regarded as an important determinant.However, while the theoretical pathways through which infant and child mortality affect fertility are well understood, the empirical evidence has been inconsistent. Thus, the unsettled nature of the link between childhood mortality and fertility was a major motivation for this study. Methodologically, the paper examines the usefulness of frailty models in exploring the childhood mortality-fertility relationships using DHS data from Ghana and Kenya. Invariably, women with prior infant deaths were found to have more subsequent births than those without mortality experience, suggesting both a physiological and behavioral response. While corroborating this finding, the multivariate results demonstrated that models without unobserved heterogeneity tended to produce biased estimates. Comparing Ghana and Kenya, there were significant differences in the effects of childhood mortality on subsequent births. At all parities, the fertility response to mortality was found to be larger in Ghana, perhaps suggesting a negative relationship between fertility response and the stage of fertility transition.