Levels of childlessness reported by the same cohorts of women and surveyed by the British General Household Survey increase with age by over 50 per cent between their early 40s and their late 50s for women born in the early 1940s. Yet the reported mean fertility of parous women remains constant with increasing age. Similar results hold within sub-groups such as different educational and marital status groups. Possible reasons for these findings include declining comparability of the base population over time due to differential migration, mortality, or institutionalization, changes in survey design and content, and differential non-response by childless and parous women. The evidence shows that these are not adequate explanations. The most likely reason is deliberate under-reporting of childbearing among older cohorts of women. The findings have implications for the interpretation of retrospective data in surveys and for the availability of informal care for older people.