Studying　the　topology　of　a　rail　transit　network　based　on　complex　network　theory　is　of　great　significance　for　identifying　the　weak　links　of　the　network　and　improving　the　network＇s　accessibility　and　connectivity.　Transfer　stations　play　important　roles　in　the　network.　This　study　investigates　the　dynamic　properties　of　the　transfer　network　of　the　Beijing　rail　transit　system　considering　the　passenger　flow　assignment.　The　P-space　representation　and　automatic　fare　collection　data　were　employed　to　build　a　directed　and　weighted　network.　The　improved　degree　centrality,　closeness　centrality,　betweenness　centrality　and　the　PageRank　index　were　used　to　distinguish　the　key　stations　and　sections　of　the　network.　A　community　detection　method,　the　Infomap　algorithm,　was　applied　to　partition　the　network　into　some　subnetworks　according　to　the　dynamics　of　passenger　flow.　The　results　showed　that　a　large　majority　of　the　important　stations　are　in　large　residential　areas　or　commercial　office　areas,　located　at　the　intersections　of　the　radial　lines　with　the　loop　lines.　According　to　the　travel　demand　of　passengers,　the　network　was　divided　into　8　communities.　The　majority　of　the　trips　start　and　end　in　the　same　community.　This　study　could　help　effectively　identify　the　important　stations　and　sections　as　well　as　the　community　structures　generated　by　passenger　mobility.　(C)　2019　Elsevier　B.V.　All　rights　reserved.