Prospective models of intergenerational reproduction consider not only the social reproduction of attributes such as education, but also the demographic reproduction of individuals who carry these attributes. This study introduces a method that substantially reduces the data requirements of prospective models. The method allows estimating prospective models based on retrospective data commonly available in surveys. It addresses two key issues – retrospective sampling bias, and the problem of identifying representative cohorts. Additionally, we introduce a decomposition technique disentangling social and demographic pathways of reproduction. We illustrate the method using German data from NEPS and SOEP, focusing on educational reproduction of German men and women born between 1930 and 1945. The analysis offers new estimates of the expected number of higher and lower educated children born to men and women of different levels of education. Findings show that reproduction is constrained by educational gradients in women’s fertility, particularly in West Germany. We conclude that prospective designs advance our understanding of social inequality and its reproduction in families. The method that we introduce will facilitate future prospective research on social stratification.