Saturation pressure is a crucial property of reservoir fluid, which is widely used in petroleum engineering calculations. In this study, a neural network model was developed for prediction of saturation pressure based on molecular components of 100 crude oils samples. Inputs of a neural network model include mole fractions of hydrocarbon and non-hydrocarbon components (C1C6, C7+, H2S, CO2, and N-2), specific gravity of C7+, molecular weight of C7+, and reservoir temperature. Molecular components of 30 crude oil samples were employed to judge the efficiency of the proposed neural network model against the Elsharkawy model and SoaveRedlich-Kwong and Peng-Robinson equations of state. Results showed that the neural network model achieved the lowest average absolute relative error and the highest correlation coefficient. Furthermore, relative contribution of network inputs on saturation pressure was obtained by sensitivity analysis, which indicated high influence of C-1 and H2S mole fractions on saturation pressure.