In this study, we demonstrate the importance of incorporating shippers’ preference heterogeneity into the optimization of the China Railway express network. In particular, a bilevel programming model is established to minimize the total construction cost for the government in the upper level and maximize the shippers’ satisfaction in the lower level. The proposed model considers price, time, reliability, frequency, safety, flexibility, traceability, and emission. Two designs are obtained by applying the model to two scenarios, in which one is of the aggregate shipper group and the other is of the three distinct clusters. Results show that explicitly including heterogeneity in network optimization pays off in terms of the dramatic increase in shippers’ satisfaction and the share of the sustainable railway without generating extra cost for the system. The results of this study could lead to insightful implication for proper network planning for the China Railway express and some useful suggestions on the subsidies of the government.