Abstract: In order to predict the residual life of corroded oil and gas pipelines in service, a model for predicting the residual life of pipeline based on the modified Generalized Extreme Value(GEV) distribution was established after the GEV adaptive optimal distribution was investigated in this paper. First, the parameters of GEV distribution function were estimated by means of Markov Chain Monte Carlo method(MCMC)and the type of extreme value distribution was determined accordingly. If the graph is inspected reasonably, this distribution will be used to predict the maximum corrosion depth. Then, the corrosion allowance prediction model was established based on the reliability and safety of pipeline.Finally, based the pipeline data, e.g. maximum corrosion depth, corrosion allowance and service life, the index model of the relationships among them was established to predict the residual life of pipeline. Furthermore, a domestic oil and gas pipeline was taken as the research object, and its residual life was predicted. It is indicated that this model has a high prediction precision and is not limited to the specific distribution of data. And as the prediction model for the residual life of pipeline, its universality is better.