Experiments were conducted to evaluate the effect of road surface characteristics on the rolling resistance of two passenger cars as part of a program to quantify vehicle operating costs as a function of road condition. The coast-down technique was used and proved to be practical, producing repeatable results. Regression models were developed to predict the rolling resistance coefficients in terms of factors such as tire temperature, road surface texture and road roughness. A comparison of fuel consumption obtained by dynamometer simulation with a simplified prediction model illustrated that the latter would be adequate for planning purposes, but for detailed fuel consumption studies a simulation technique incorporating an engine torque/fuel consumption map would be required.