We utilized genetic programming approach for generation of a new model for prediction of critical properties and acentric factor of petroleum fractions. Usually for estimation of critical properties, molecular weight, specific gravity and boiling point are needed, while in this work in order to present an applicable model, just the boiling point and specific gravity have been used as the model parameters. Our results showed that using new correlations, critical temperature, pressure, volume and acentric of petroleum fractions can be correlated within 0.07, 0.65, 0.45 and 0.79 percent average absolute relative deviation. respectively.