For each quartile of income per household or per consumption unit, annual time-series have been estimated from panel surveys, with annual waves of observations from 1974 to 2007: INSEE2 Households' "Conjoncture" survey from 1974 to 1994 ; panel "Parc-Auto"3 Sofres4 since mid-80's. In these data sources, household behaviour is described through ; car ownership (percentage of households with at least one car, of which percentage of multi-car households, average number of cars per adult over 18, which is the minimum age for driving license in France) ; car use (annual mileage per household or per car). The repeated sample structure of data has been used for improving the accuracy of time-series of variables highly correlated for subsequent years [Cochran, 1977]. In mid-70's, car ownership and use were quite low for the poorest income quartile, but the difference has much decreased with all the three higher income groups, which are more homogeneous. Thus, multi-car ownership, which is mainly structured by geographic and demographic determinants, has slowed down -but not reversed- the social diffusion of automobile. As the curves representing car ownership (number of cars per adult) and car use (annual mileage per household) seem to become quite horizontal during the most recent period, logistic curves have been estimated according to time, then to real income. For each quartile of the distribution of households by income per consumption unit, saturation thresholds are estimated, as well as the date or the point of inflection. Follows a discussion on the legitimacy of pooling the data. The relationship between temporal elasticities (for each quartile) and cross-sectional income elasticities, which can be considered as a measurement of inequality at each point in time, will be discussed [Gardes and Madre, 2005].