An understanding of the errors found in census questions is important in order to assess the level of confidence in the census data and to get an appreciation of the impact of using these data in estimating mortality derived from census data. While demographic methods are often used to determine the direction of bias in the data, direct evaluation studies are required to determine the nature and extent of biases with more accuracy. Equally important is an understanding of the characteristics of the respondents who produce better responses in censuses. This can be used to inform selection of better respondents in order to improve the quality of the collected data. This research uses census data from a survey, which is matched to the longitudinal Health Demographic Surveillance System site (HDSS) data from Agincourt, Limpopo Province, South Africa, in order to assess the biases found in data used in child and adult mortality estimation that uses indirect techniques.