As a universal basic component of society, the family is a determinant in all socio-demographic research issues, whether in fertility or health, education, gender relations, family economy, or ageing. In the absence of specific sources, quantitative family studies are largely based on secondary analysis of census data, or data from other cross-sectional surveys. For statistical purposes, the family is thus studied from the viewpoint of household composition whereas there is not always a one-to-one relationship between households and families. The aim of this manual is to suggest a method for reconstructing the family composition of households using sources that are widely available in the South, and are both regularly renewed and accessible, namely the Demographic and Health Surveys (DHS). The household datasets in DHS surveys are not very complete from the point of view of relationship to the head of household. However, they do contain additional information concerning persons eligible for the individual survey, i.e. women aged between 15 and 49 and children under 15 years. By using these variables it is possible to identify the various family nuclei that make up a household. This method of using all the available – albeit fragmentary – information, allows us to identify cohabitation of family nuclei and also to determine the nature of secondary units, whether these are composed of a couple with or without children or a single relative with children, as well as their relationship to the head of household (collateral units, ascendants, descendents, non-related and so forth). This approach considerably enriches household analysis and enables us to determine individual positions within the household by specifying the residential status of respondents. The authors present an application to the study of the family composition changes in Cameroon households between the 1991 and 2004, using DHS data.