Abstract: Several cases of failures in steel buried pipelines under the effect of wave propagation have been reported. Due to seismic waves propagations these pipelines will encounter various axial forces and bending moments which will consequently lead to the local buckling of the pipes and the reduction of the pipes hollow-sectional area. These effects cause overall reduction of efficiency of pipes. Due to the probabilistic nature of soil and earthquake specifications, a deterministic approach for analyzing buried pipeline networks against earthquake excitations is not appropriate. In this study an algorithm for reliability assessment of buried pipeline networks is proposed which is based on nonlinear dynamic analysis and calculation of reliability using Monte Carlo simulation. Due to complexity of numerical analyses of buried pipeline networks, there is no possibility of an explicit calculation for the performance limit state function, so a trained multilayer feed forward neural network was used as an alternative. For this purpose, the obtained results of many deterministic numerical analyses were used for training the neural network and the performance limit state function was replaced by trained neural network. Finally, based on the probability density function, standard deviation and average of probabilistic parameters, reliability of the pipeline network for different performance levels, was determined. By investigating a buried pipeline network in sandy soil as a case study, effectiveness of the proposed algorithm was investigated and by determining the importance measure of probabilistic parameters, sensitivity analysis was performed.