The composition of households frequently change due to births, deaths, divorces, marriages, the departure of children from home, and other compositional changes. Consequently, a large number of people undergo some fundamental change in household arrangements during relatively short periods of time. However, using data from Indonesia, this study finds that change in household composition is not a major cause of chronic poverty. Similarly, it finds no evidence that households change their composition to cope with negative shocks. Nevertheless, the study confirms that the larger the number of household members, the higher the probability that a household is chronically poor. Comparing different types of household compositions, households with a single female without children have the lowest probability of being either chronically poor or vulnerable, while single male households with or without children have the highest probability of being vulnerable. Frequent changes in household compositions imply that the use of household as the unit of analysis for poverty may undermine, or at least complicate, the conceptualization and measurement of chronic poverty. This also implies that the problem of targeting social protection programs not only relates to implementation, but also has some conceptual roots.