<p>A number of studies have shown that fertility levels differ substantially across settlement types in modern societies. As a rule, as the population density increases, fertility decreases. Such differences can occur due to a combination of three factors: (1) direct contextual influence, (2) differences in population composition between settlement types, and (3) selective migration. In this study, we aim to disentangle these factors with respect to completed fertility. We use linked micro-data from the Estonian population and housing censuses of 2000 and 2021 that we analyse with Poisson regression. The results show large fertility differences between settlement types, with women living in the capital city having the lowest, and women living in the more distant countryside having the highest number of children. Control of individual socio-demographic and housing-related variables (size, type, ownership), as well as migration experience, somewhat reduces the effects of settlement type. The analysis underlines the size of the dwellings to be especially relevant for the contextual effect. Looking at intercensal migration flows in greater detail reveals that the completed fertility of migrants lies mostly between the origin and destination groups.</p>