We use core analysis and well testing to determinate the reservoir lithology. Unfortunately, coring from each wells in large oil fields such as Iran oil fields, is very expensive. However, because of the importance of this information which is obtained from lithology, it is necessary to coring from some of the reservoir wells.Purpose of this study is prediction of hydrocarbon reservoir lithology in South Pars field using artificial neural network with back propagation error algorithm (BP) and Trainlm algorithm with Matlab software from wire-line logs including gamma ray, density, neutron, sonic and photoelectric (PE). This method can reduce requirement of coring and reduce the costs. The area we have studied, consist of three lithologies, including Dolomite, shale and Anhydrite. The regression between the predicted and the real values of volume concentrations of Dolomite, shale and Anhydrite are obtained respectively, as 0.87, 0.76 and 0.90. The results show that the neural network gives a reasonable estimation for lithology.