Petroleum consumption of vehicles is always a concern of many prospective car buyers, yet many seem clueless about its determinants. This paper determines the main predictors of petroleum use of vehicles from six predictor variables, and also developed a statistical model for their consumption. Data were collected on 121 common vehicles in Ghana from seven automobile companies in Accra, Ghana. Statistical analyses used included Pearson's correlation and Best Subsets Regression, where the Mallow's Cp and adjusted R-square helped to determine the most adequate and plausible regression model. Key findings showed that weight, width, and horsepower were the highly, negatively and significantly correlated with petroleum use of vehicles. However, the best regression model for prediction involved only Weight and Horsepower. The study bears implications for prospective vehicle buyers, distributors and the Government.