Abstract: 37 gasoline blending components and hydrocarbons composition of refined oil are analyzed by GC/MS, including nalkane, isoparaffin, naphthenic hydrocarbon, aromatic hydrocarbon, isoolefine, normal olefin, cyclic olefins, diolefin. The author adopts correlational analysis method to investigate the relativity between hydrocarbon substance and gasoline property; adopts stepwise linear regression method to establish a regression model of property and hydrocarbons composition; and determine the main hydrocarbon substance which exerts influence on the property according to the regression model: aromatic hydrocarbon influences the octane number, 10% vaporization temperature, 50% vaporization temperature, density and chroma most; olefin, especially the isoolefine and normal olefin, influence 90% vaporization temperature, acidity, colloid, induction period, iodine value and saturated steam pressure most. F test result shows that the regression model is obvious and it can provide reference for predicting the quality of gasoline.