This paper describes research on the demand for gasoline by automobile drivers. It discusses the relationship between the ownership of and the use and fuelconsumption of automobiles. In view ofthedifficultyinrelating behavioral hypotheses about individuals and households to aggregate data, the intricacies of the new-and used-automobile markets are presented. Aggregate gasoline demand models are reviewed and, where available, short-run price elasticities of gasoline are given. Variables, functional forms, and levels of aggregation are indicated. A method of integrating time-series and cross-sectional automobile data and a hypothesis about the prices of services of different sorts of automobiles are discussed. Two other models that simultaneously treat the demand for automobiles and gasoline are reviewed: They are based on (a) the different size classes of new automobiles and aggregate automobile travel as the jointly dependent variables and (b) the new-and used-car markets and aggregate automobile travel as the interrelated entities. These models used only annual data at the national level. Our empirical analysis consists of a single equation model for which the dependent variable is per capita gasoline consumption. The predetermined set includes a lagged dependent variable, demand for new automobiles, deflated gasoline price, and gasoline consumption per automobile at the annual and national levels. Some alternate forms of the hypotheses are given, and the results of estimation are presented andcompared. The most reasonable specification produces a short-run gasoline price elasticity estimate of-0.23, a result midway among those of other investigators who have based estimated elasticities on similar data sets.

• ONLY recently has there been interest in modeling gasoline consumption as a consumer product. Most attempts to model gasoline consumption have ignored or have lightly treated possible adjustments in ownership, purchase, and use of automobiles. Similarly, past attempts to model automobile ownership or purchases over time (or over cross sections) have mostly ignored the influence of gasoline price or gasoline consumption as determinants.