Fertility behaviour in France and Belgium in the first half of the twentieth century is still quite poorly understood, owing mainly to legislation that prohibits the use of individual data less than 100 years old, and to the paucity of cross-sectional aggregate data. This article tests whether it is possible to bridge that gap via a retrospective approach, using the questions about women’s reproductive life from family surveys conducted in France and population censuses in Belgium. The analyses show that the risks of bias related to recall errors, selection effects and non-response, inherent in these retrospective observations, are limited. The reconstituted birth histories of the various cohorts of women recorded at the time of the survey or census therefore offer considerable research potential, making it possible to calculate more detailed, more diversified indicators of the intensity and timing of fertility for cohorts of women born since the end of the nineteenth century. They also enable us to capture the spatial dimension of fertility, at the level of the regions in France, and of the arrondissements and even the municipalities in Belgium.