The present work develops a computer-aided algorithm to transform gasoline bulk property into molecular composition. We pre-defined 166 representative gasoline molecules according to the compositional characteristics. Simulated annealing method was used to find a molecular distribution that has bulk property very closed to the measured data. The accuracy of the model is verified by using a set of FCC gasoline data. The relative error of predicted and measure key property is lower than 2% and the overall relative error is lower than 5%. The predicted PIONA composition is in good agreement with experimental results, indicating that the obtained compositional information is in consisted with the actual composition.