Ticket price of high-speed railway in China is unitary and changeless, which cannot adjust passenger flow and improve revenue. Based on the analysis of passenger's sensitivity towards ticket price in different booking periods, a robust optimization model of multi-period dynamic pricing for high-speed railway was developed aiming to maximize the revenue on the basis of revenue management theory and the transportation organization characteristics of high-speed railway. Considering the uncertainty of demand function, the model was solved with a robust optimization method. Beijing-Shanghai high-speed train was taken as an example to test the effectiveness of the model. The results showed that compared with the seat allocation scheme under the unitary ticket fare, the proposed model can dynamically adjust ticket fare to regulate the passenger flow in different booking periods and improve revenue. In addition, the robustness of the model was verified by the simulation test. Compared with the nominal model, the proposed model can better adapt to the fluctuation of the demand and ensure that more tickets are sold at a higher price. The proposed model provides an optimization method for dynamic pricing of high-speed railway.