Identification of the gasoline purity is important for quality control and detection of gasoline adulteration. Principal component analysis and Raman spectroscopy were used to authenticate gasoline adulterated with methyl tert-butyl ether (MTBE) and benzene. Gasoline could be clearly distinguished from gasoline adulterated with MTBE and benzene by a plot of the first principal component (x-axis) against the second principal component (y-axis). And the radial basis function neural network was used for quantitative prediction of the volume percentages of MTBE and benzene in gasoline based on Raman Spectra. The correlation coefficient (r) and mean absolute percentage error between predictive values and spiked values were 0.9907 and 0.9934 and 15.73 and 8.19%, respectively. Moreover, the Raman spectra of the samples were obtained with a portable Raman spectrometer. Therefore, the method is simple, effective, fast, does not require sample pre-processing, and is promising for rapid gasoline detection. Copyright © 2012 John Wiley & Sons, Ltd.