The segregation–crime relationship is a classic topic in sociology and crime geography, yet existing literature mainly focuses on the impact of racial segregation at the global scale. Little is known about the impact of local segregation of other socioeconomic characteristics such as education level, an important segregation factor for racially homogenous countries like China. Also unknown is their impact beyond the residential domain. Using the Baidu Map Location-Based Service population data set and court records in 863 local geographic units of the central urban area of Beijing during 2018 and 2019, this study uncovers the spatial pattern of segregation between people with and without a bachelor’s degree measured in the residential space and activity space and further investigates the influence of these two types of educational segregation and their interaction effects with social context on theft and violent crime. Results show less segregation in the activity space than in the residential space. Both types of segregation, however, significantly increase the risk of theft and violence, with activity space–based segregation more consequential. Moreover, the positive segregation–crime link is moderated by the local social context measured by the educational composition among residents and the ambient population. Compared with residential segregation, activity space–based segregation is more detrimental for places dominated by the less educated. Our results highlight the elevated influence of segregation on safety beyond the residential space, especially for areas clustered with the less educated ambient population.