Abstract: In view of the limitations of traditional grey prediction model, which strongly depended on historical data, had small fault tolerance and poor anti-interference ability, unbiased grey theory and residual verification theory were introduced to the prediction model, and the traditional grey model was improved from the trend curve grey fitting and state classification method, the unbiased grey prediction of railway passenger traffic volume was proposed. The forecast method of railway passenger traffic volume was expounded from two aspects of qualitative prediction and quantitative prediction, and its advantages and disadvantages were analyzed. Based on the unbiased grey forecast model of railway passenger traffic volume and the data of railway passenger traffic volume in 1997-2016, the data of railway during the "13th Five-Year" period were predicted by the residual test and the result analysis. The prediction accuracy is obviously higher than that of the BP neural network prediction.