Demographic indicators such as fertility rates and infant mortality rates are often measured in census surveys by interviewing mothers to obtain their pregnancy histories and child deaths. The validity of such surveys depends upon accurate recall of histories, truthful reporting of events and understanding of the questions posed. To measure the reliability of maternal reporting, two census surveys conducted in a rural Egyptian population were compared. Women between 15 and 55 years of age residing in 20 villages were asked their histories of live births, stillbirths and child deaths. An identical set of questions was posed 2 years later. Twice-monthly home visits were conducted in the intervening 2-year interval to identify accurately any new births, stillbirths and deaths occurring in the population. The maternal reports from the first census were combined with the prospectively identified births, stillbirths and deaths and compared with the maternal reports from the second census. For 1502 women, the discrepancies in the total number of births, stillbirths and child deaths reported between the two surveys were 0.6%, 4% and 0.6% respectively. However, when the consistency of responses was analysed, the proportion of women with discordant responses was 10%, 6% and 7% for the same measures. These results suggest that, despite the large number of births and deaths that women may experience in developing countries, maternal interviews provide reliable responses that can be used to estimate mortality and fertility rates in settings where vital records are incomplete or unreliable.