Abstract: In view of the complexity and uncertainty of the factors contributing to the gathering pipeline corrosion and the difficulty of predicting the remaining service life of corroded pipe, two methods, the probabilistic statistics and artificial nerve network, are compared and studied to predict the remaining life of in-service corroded pipe.These methods can predict the corrosion variability of pipeline and provide the remaining life predicting credibly.Finally, the authors prospect the developing trend on predicting methods.