As one of the major means to link refineries to local markets, pipelines are crucial for refined oil supply chains. Pump scheduling is vital to optimally operate the refined products pipelines. Existing studies mostly quantified the number of pump stop/restart in the form of cost, whereas the cost coefficient was often subjectively selected. The present study built a multi-objective mixed-integer linear programming (MOMILP) model for refined products pipelines to minimize the number of pump stop/restart and reduce the pump running cost simultaneously. In this study, the minimum continuous running time of pumps, changing electricity price (i.e., time-changing electricity price and local electricity) and the pressure limits of special points were considered meticulously, e.g. high-elevation points, low-elevation points and pump stations. In the first stage, the improved augmented epsilon-constraint method (AUGMECON) was adopted to deal with this model, and obtained a seris of pump operating schemes, acting as a Pareto set. Then the results of the AUGMECON were evaluated by an established evaluation model based on neural network. Finally, two cases from a refined products pipeline system in China were employed to verify the practicability and accuracy of the proposed model. The results of this study can effectively guide the pump scheduling of refined products pipelines. (C) 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.