AbstractThe successful development of bicycle-sharing systems (BSSs) is influenced by the socioeconomic characteristics and geographic attributes of metropolitan areas. Trip generation and attraction volumes, which represent the actual flows of resident activity in a specific area, may influence BSS ridership, particularly for travelers using a BSS for first- or last-mile services. However, most studies have used population and other socioeconomic data to investigate BSS ridership without considering trip attributes. Population-related attributes may influence BSS ridership, but they cannot account for the spatial distributions of vehicular or passenger trips between specific origin–destination pairs. In contrast to past studies, this study collected 9 years of BSS rental data and related socioeconomic characteristics of the CityBike system in Kaohsiung City, Taiwan, including residents’ trip attributes. Panel data were analyzed using autoregressive with exogenous variable models. The results indicated that trip attributes that are more appropriate than population data are in BSS ridership prediction. Accurate predictions of BSS ridership volumes over time enable the allocation of limited resources to establish new stations or infrastructures for sustainable BSS development.