A two-dimensional numerical model, involving energy conservation, momentum transport and continuity equations, was developed to provide the profiles of temperature, velocity and pressure of vacuum membrane distillation (VMD) process using hollow fiber module. The theoretical prediction was in good agreement with the experimental results from literature. Four design and operation variables, including feed temperature, hollow fiber length, feed volume flow rate and vacuum pressure, were optimized to minimize water production cost (WPC) using genetic algorithm (GA). A case study shows that the WPC can be decreased by 38.1% in comparison with the non-optimized VMD process. Meanwhile, a general guidance for reducing the WPC was also provided. The feed temperature should be adopted as high as possible, while the hollow fiber length cannot be too long to result in a decrease in water flux and an increase in pressure drop. The flow rate is recommended to take a lower bound value and the moderate degree of vacuum is preferable. This study indicates the importance of VMD system optimization to minimize water production cost. (C) 2014 Elsevier B.V. All rights reserved.