The　literature　on　segregation　is　focused　on　the　residential　domain　considered　from　a　static　perspective.　In　contrast,　the　purpose　of　our　study　is　to　examine　temporal　variations　in　the　overall　degree　and　spatial　pattern　of　activity-space-based　social　segregation　around　the　clock　on　weekday　and　at　weekend　in　the　central　urban　area　of　Beijing,　China.　Drawing　on　location-based　service　(LBS)　big　data,　we　measure　the　level　of　activity-space-based　segregation　at　each　hour　of　a　weekday　and　a　weekend　day　between　groups　of　people　who　differ　from　each　other　in　relation　to　formal　educational　achievements.　Their　comparisons　with　the　segregation　in　major　life　domains　such　as　residence　and　workplace　are　also　made.　At　the　global　level,　the　extent　of　activity-space-based　segregation　fluctuates　around　the　clock,　with　less　segregation　during　the　daytime　than　at　night　and　less　segregation　on　the　weekend　day　than　on　the　weekday.　The　segregation　degrees　for　all　groups　are　in　descending　order　workplace　segregation,　residential　segregation,　and　out-of-home　non-employment　segregation.　At　the　local　level,　the　highly　segregated　units　centralize　to　city　center　in　the　morning　and　decentralize　to　suburban　areas　in　the　evening.　The　spatial　segregation　patterns　at　various　times　of　the　day　change　to　a　much　greater　extent　on　the　weekday　than　during　the　weekend　day,　especially　for　employment　centers　and　large-scale　residential　communities.　Lastly,　a　spatial　unit　classification　framework　of　real-time　activity-space-based　segregation　is　proposed　to　integrate　multiple　kinds　of　information　pertaining　to　the　segregation　level　and　the　dominant　group　in　a　given　area　at　a　given　time　with　the　extent　and　trend　of　the　temporal　variation　identified　presented　as　a　concise　map　useful　both　to　advancing　further　research　and　guiding　policy　formulation.