During　the　period　of　2013-2015,　a　total　of　34　PM2.5　pollution　episodes　occurred　in　Beijing,　each　of　which　remained　for　at　least　2　days.　Among　that,　28　times　occurred　in　winter　half　year　with　the　average　concentration　of　243.1　mu　g/m(3)　and　summer　half　year　with　the　average　concentration　of　194.1　mu　g/m(3).　These　episodes　were　mainly　associated　with　lower　wind　speed　and　lower　visibility　as　well　as　higher　relative　humidity,　indicating　that　they　belonged　to　heavy　pollution　under　static　stability.　The　PM2.5　pollution　was　classified　into　two　categories　according　to　the　back　trajectory　analysis　and　meteorological　background　field.　Category　1,　accounting　for　22　times　among　all　the　pollution　episodes,　was　due　to　air　mass　transport　from　Beijing＇s　southern　regions　with　north-south　direction　pressure　gradient　and　sparse　isopiestic.　And　category　ll　was　mainly　led　by　northwestern　air　masses　accompanied　with　a　large　area　of　uniform　pressure　field.　Then,　a　typical　case　study　was　conducted　for　each　category　to　recognize　the　sub-region　contribution　to　Beijing＇s　PM2.5　pollution　based　on　WRF-CAMx　modeling　system,　and　the　simulation　results　indicated　that　local　emission　source　contribution　decreased　significantly　during　the　accumulation　phase　for　category　I,　but　increased　during　that　of　category　II,　with　an　average　contribution　of　47.3%　and　77.1%　during　the　entire　pollution　period　of　each　category,　respectively.　Two　red　alerts　of　air　pollution　occurred　in　December　2015　were　also　analyzed　based　on　the　episode　classification.　It　was　found　that　the　second　red　alert　pollution　episode　belonged　to　category　II.　The　emission　control　measures　in　Beijing　worked　more　obviously　with　the　reduction　effect　ratio　of　15.4%　compared　to　the　first　red　alert　period　(9.7%).　(C)　2017　Elsevier　B.V.　All　rights　reserved.