Nowadays the society and the legal institutions undertake efforts to minimize the CO2 emissions to the atmosphere, avoiding the global warmth. For the automotive industries, it is not different. It is necessary to develop vehicles with more energetic efficiency, that can meet a bigger range with less fuel spent, emitting less CO2. An important step on a vehicle development is the earlier determination of fuel economy (FE), depending on the projected vehicle characteristics and depending on the vehicle fuel economy market positioning. Several ways for calculate FE were developed along the time, including hybrid regression models, using key input variables like instantaneous vehicle speed and acceleration measurements, theoretical approximation methods derived in terms of physical properties of engine-vehicle systems, relationship among traffic related parameters, power-demand models, artificial neural networks and genetic algorithm, all of them using vehicle input data like engine speed, torque, fuel flow, intake manifold mean temperature, or make of car, engine style, weight of car, vehicle type and transmission system type, and so on. However, there is not in the literature an easy and quick way to make FE estimations of a vehicle. in a context of a group of similar vehicles, which is the purpose of this work. The proposed empirical physical model delivers FE estimations for a vehicle using minimum relevant input information: engine displacement, performance (recovery speed times) and road load data, through mathematical adjustment of parameters. The empirical model is applied in a Brazilian study of case for validation and the results indicates the success of the method for city cycle and highway cycle, within a 6% margin of error, comparing with the official Brazilian government public data for this tested vehicle.