Urban　agglomeration　is　an　important　strategy　used　to　promote　economic　development　and　urbanization　in　China.　Understanding　the　structure　of　urban　agglomeration　is　therefore　essential　for　policy-makers　and　planners.　In　this　study,　the　Beijing-Tianjin-Hebei　urban　agglomeration　(BTHUG)　is　explored　through　a　proposed　spatial　network　analytical　framework　and　a　large　mobile　phone　data　set　(over　20　million　users).　We　first　construct　a　weight-directed　spatial　interaction　network　based　on　an　origin-destination　matrix　derived　from　the　data　set.　Several　network　metrics　(i.e.,　degree,　strength,　the　rich-club　coefficient,　and　the　assortativity　coefficient)　and　three　selected　community　detection　algorithms　(i.e.,　Infomap,　Louvain,　and　Regionalization)　are　applied　and　compared　to　reveal　the　structure　of　the　BTHUG.　A　four-level　hierarchical　structure　is　defined　and　observed:　one　global　center,　two　local　centers,　major　cities　that　have　low　mobility　flow　but　strong　linkages　with　the　three　centers,　and　peripheral　cities　that　have　low　mobility　flow　and　weak　linkages　with　the　three　centers.　In　particular,　the　results　imply　that　the　spatial　structure　of　the　BTHUG　is　over-dependent　on　the　global　center　(i.e.,　Beijing　and　northern　Langfang).　Further,　ignoring　spatial　interaction　patterns　in　top-down　administrative　planning　for　urban　agglomeration　may　lead　to　ineffective　integrated　development.　The　implications　for　BTHUG　planning　are　discussed.