The collection of data on the sex classification of household members in surveys and particularly in World Fertility Surveys (WFSs) is discussed in regard to why and how the data is collected and how reliable the data is. The degree to which results may be biased by the different responses to survey questions made by male and female respondents is also discussed. In surveys the sex of the respondent is simply recorded by the interviewer; however the respondents are generally asked to provide information on the sex of other members of the household. Sex classification data is sometimes used to assess social and cultural differences between men and women. In demographic studies many variables are analyzed by both age and sex as there are often marked differences between males and females of similar age in regard to education literacy occupational status family roles mortality and many other factors. Sex reporting tends to be more accurate than the reporting of most other demographic variables but it is not totally accurate. Interviewers may guess rather than ask the sex of household members who are not present. Respondents themselves may misreport the sex of infants who died shortly after birth. Data processing procedures may result in sex misclassifications. According to the processing guidelines used in some WFSs males are coded as 1 and females are coded as 2 but all unknown responses are also coded as 1 and are thus indistinguishable from males. Sex classifications are sometimes made simply on the basis of occupation. Often the head of the household is automatically assumed to be the male in households which contain both a man and a women. Sex differences in reporting makes it necessary to carefully consider whether respondents in a study should be restricted to 1 sex only or should include both sexes. Males are less likely than women to be able to report accurately on breastfeeding and on the fertility behavior of household members. On the other hand in some societies males may have more control over family size than women. In these societies conclusions made on the basis of data on family size ideals collected from women only may be faulty. Respondents also tend to be more accurate in reporting about themselves than in reporting about others. For example economic activities of female household members are frequently underreported by male respondents and females frequently underreport vasectomies of male partners.