As an important component of multimodal public transport, bus-metro transfer is a key link in the integration of urban passenger transport. Based on multi-source data, this paper constructed a multi-scale geographically weighted regression model (MGWR) with bus-metro transfer ridership as the dependent variable during the weekday morning peak period in Beijing to reveal the influence and spatial heterogeneity of traffic transfer conditions, the level of development around metro stations, and metro station characteristics on transfer ridership. The results showed that MGWR has stronger explanatory power than linear regression model(OLS) and geographically weighted regression model (GWR), and the influencing factors of transfer had significant spatial heterogeneity; the number of bus routes, microcycle bus routes, bus stops, the number of hub and tourist POIs, and the mixed index of land use had a facilitating effect on transfer; the average walking transfer distance, road density, and local station density had a suppressing effect. This study provided evidence to reveal the influence mechanism of bus-metro transfer ridership through more comprehensive and detailed factors, which was important for promoting a more rational and effective integration of the public transport and improving commuting conditions.