This paper studies poverty as a dynamic phenomenon, motivated by the recurring economic crises that affect developing countries and the incidence of income fluctuations on household welfare. While the increasing availability of household panel data has been exploited in theoretical analysis and empirical applications, the methodological and applied literatures still lack a unified framework. Echoing Atkinson (1987), this paper addresses the question of how poverty should be measured over time – or, in more general terms, how to measure well-being based on repeated observations of household income. The paper develops and illustrates a set of tools for empirical work based on theoretically sound extensions of the existing methodology for static distributional analysis. Moreover, this framework encompasses some of the existing approaches as special cases. These tools are illustrated with longitudinal data for Argentina in the 1995- 2002 period, which is well suited for this type of analysis given the large fluctuations in household income due to the repeated economic crises in the country.