This paper evaluates how gasoline prices influences the average fuel econ- omy of the existing automobile fleet. Higher fuel price affects fleet composi- tion in two ways: immediate purchase decisions of new, more fuel efficient, vehicles and scrappage of old fuel inefficient gas-guzzlers. Gasoline costs ac- count for 65% of the total operating costs of driving an automobile. Rational forward-looking consumers will account for both current and expected future gasoline prices to decide not only what vehicle to purchase but also when to purchase it. Scrappage of old cars will also be driven by the same considera- tions, plus their increasing maintenance cost, and improved features of new models. In order to account for all these dynamic effects on the composition of the automobile fleet I specify and estimate a structural dynamic model of consumer demand for new and used vehicles as in Gowrisankaran and Rysman (2009). However, my model not only predicts the market shares of each vehicle sold in every period but also the survival probability for each model-vintage for each sample period. I estimate the model using a rich dataset combining vehicle registration and current fleet composition of sev- eral cities between 2003 and 2009 that include vehicle characteristics, price, gasoline price, and demographics for all market-years. Parameters are esti- mated by matching the predicted market shares and survival rates of every model-vintage with the corresponding empirical moments over the time span of the sample. Parameter estimates are then used to evaluate substantial fuel tax increases that have never been implemented before for being considered controversial and/or politically risky. Preliminary results for the Houston and San Francisco markets indicate that a permanent increase of gasoline price to $4 per gallon has stronger (and stable) long term effects than just doubling the current gasoline tax, which leads only to a temporary increase of the average fuel economy of the automobile fleet.