Several empirical studies have established that tenure choice and households mobility decisions are highly correlated (Ozyilirim et al., 2005). In the literature, there have been two methods that have used to estimate tenure choice models. The first uses a sample of recent movers while the second employs a sample of all households. Other approaches are mixed (Painter, 2000). This work refers on the results of an empirical analysis, developed to the urban level, by a sample of movers (renters and homeowners). The main goal is to test if homeownership is systematically preferred to leasing and if the housing affordability effect can be estimated by a tenure choice model. An econometric model (Logit) is used to this purpose; it is normally employed in the study of binary choices but here it is also estimated to the policy aim. The analytical results confirm, as expected, the strong trend to the homeownership for the families of renters. But, in current econo mic situation, the high value of the rent is systematically associated with a reduced chance of changing for the medium-income families and this do not concern the low-income households. That situation has important effects on the social policies as the households mobility choices in the housing market produce effects that influence the urban traffic conditions, the patterns of poverty concentration and ethnic segregation, the quality of public schools, the access to economic opportunities, and, in other hands, the decline or revitalization of the neighborhoods ó all of which are important, long-term, and related policy concerns.