A Raman spectrum fitting method based on Voigt function was proposed. The method can be used to fit the profile of Raman spectrum produced by unknown components in sample based on Voigt functions. In the experiment, pure methanol was added by various volume fraction ( 2. 5% -80. 0%) into three base-gasoline with different compositions. It can be applied to fit the background Raman spectrum produced by basic-gasoline comprised of unknown compositions. The remained spectrum, after deducting the fitted spectrum from the mixture spectrum, was considered as the contribution attributed only to methanol. And then a linear calibration model was built based on 4 training samples to predict the concentration of methanol. The root mean square error of prediction (RMSEP) was 1. 86% (V/V) and the correlation coefficient (R2) was 0. 987. Results show that it is an effective method to solve the problem of strong spectral overlap in mixture, and it has the advantages of high generalization and few training samples.