Sensitivity method used for reducing the chemical reaction kinetics model is introduced. In this method, with the aid of principal component analysis of the matrix F~TF where F is the rate sensitivity matrix, relatively important reactions can be identified from the complex reaction system by ranking the significant eigenvector elements of reaction subgroup with large eigenvalues of matrix F~TF. This method is further applied to analyze two chemical reactions, i.e., the high-temperature formaldehyde oxidation reaction and the carbon monoxide/hydrogen/air combustion reaction. The results show that both reactions can be significantly reduced by this method and the calculated species concentrations from reduced mechanism agree well with those calculated from detailed mechanism.