Abstract: Entering the post-pandemic era, railway passenger flow is gradually rising, but there is a large fluctuation. Facing the task of improving the quality and operation efficiency of railway, accurate prediction of passenger flow is becoming more and more important. In this paper, XGBoost model was adopted for passenger flow forecast with COVID-19 pandemic, weather, and date attributes as influencing factors. Meanwhile, passenger flow data of Shanghai Railway Station from January 1, 2016 to July 27, 2020 were selected as training set and validation set and the optimal parameters of the XGBoost-based passenger flow forecast model were obtained by using 5-fold cross-validation and Grid Search. Then, the passenger flow of Shanghai Railway Station from July 28, 2020 to May 17, 2021 was predicted using this model. The result of the prediction attained a fitting degree of 0.812, indicating that the overall prediction effect is good.