Crowding events, which pose tremendous pressure to city management and society safety, are a typical manifestation of anomalous human mobility in metropolitan areas. However, we are still lacking a comprehensive understanding of the anomalous human mobility in crowding events, which is crucial for preventing crowd disasters and developing sustainable cities and societies. In this study, we analyze the individual and collective human mobility patterns in crowding events using the smart card data of six million subway passengers in Shenzhen city. The discovered individual human mobility patterns reveal the underlying mechanism of crowd formation. The discovered collective human mobility patterns can be employed to anticipate crowding events, offering timely information for transportation and crowd management.