Several quantitative structure-property relationship (QSPR) models of the correlation between the structure of different alkylbenzenes and their heat capacity and standard enthalpy have been constructed. A simple set of six numerical codes was used to represent each alkylbenzene structure on the basis of its molecular formula. The QSPR models were constructed using multiple linear regression, multiple nonlinear regression and support vector machine regression. The compounds were divided into a training set and a test set. The correlation coefficient R, mean abstract error (MAE) and root mean square (RMS) of each model were calculated. It was found that the support vector machine model was superior to the other two models. The support vector machine model should also be of value in predicting the physicochemical properties of other organic compounds.