Abstract: The paper analyzed the causes of the railway passenger flow in the three-day holiday, summarized the factors of the railway passenger flow from the characteristics about passengers and travel purposes, based on the historical tickets data, analyzed the fluctuation characteristics of passenger flow in three-day holiday from transport volume, the grade of the board distance, the grade of the seat type, etc, defined the three-day holiday passenger flow fluctuation coefficient and based on a small holiday passenger trends, proposed a clustering method of time series to study the regular pattern of passenger flow in the three-day holiday and made detailed descriptions about the building of time series of the three-day holiday and its representations, eigenvalue extraction and clustering step.