The invention relates to a short-time passenger flow forecasting method suitable for an urban rail transit system. The short-time passenger flow forecasting method suitable for the urban rail transit system comprises the following steps that a, swiping card ticket data of passengers riding trail transit are obtained from an AFC; b, classification is conducted according to the types of tickets; c, analysis is conducted on different types of passenger flow data; d, according to unique identification numbers, statistics of the daily travelling rule of each one-card is carried out; e, the daily travelling rule of the single one-card in a next forecasting time period on a single day is forecasted; f, a one-card forecasting result is formed; g, for the data of disposable tickets, passenger flow distribution is conducted with a K short path method, K paths are obtained, and the passenger flow distribution probability of each path is obtained; h, the disposable ticket data of the forecasted day are distributed to the different paths according to the probabilities of the K short paths, and the passenger flow data of different paths are forecasted. Compared with the prior art, the short-time passenger flow forecasting method has the advantages of being efficient, accurate, high in real-time performance and the like.