The relationship between built environment and jobs-housing balance has been discussed at either residence or workplace in literature. Less attention has been paid to both perspectives simultaneously. Built environment may have different relationships with home-based and work-based jobs-housing balance from a single perspective. However, the difference and to what extent is still not clear. It is significant in providing area-based urban planning suggestions. This study applied one-month cellular cellphone data in Guangzhou, China. It proposed a new jobs-housing balance measurement by estimating decay parameters of commuting distance distributions. Home-based and work-based jobs-housing balance was differentiated by aggregating commuting trips into departure and arrival locations. A linear regression model was applied to examine the correlation effect of built environment factors on home-based and work-based jobs-housing balance respectively, and a classification and regression trees (CART) model was applied to explore the combined effect of built environment on jobs-housing balance patterns. Results have shown that built environment has different correlations with home-based and work-based jobs-housing balance. For example, the POI mixture improves work-based jobs-housing balance but worsens home-based jobs-housing balance. Another important finding is that a moderate jobs-housing ratio does not ensure jobs-housing balance. In contrast, jobs-housing is imbalanced at well-developed communities.