A　quick　and　accurate　identification　of　source　information　on　sudden　hazardous　chemical　leakage　accident　is　crucial　for　early　accident　warning　and　emergency　response.　This　study　firstly　regards　source　identification　problem　of　sudden　hazardous　chemical　leakage　accidents　as　an　inverse　problem　and　presents　a　source　identification　model　based　on　the　Bayesian　framework.　Secondly,　a　new　identification　method　is　designed　on　the　basis　of　particle　swarm　optimisation　(PSO),　differential　evolution　(DE)　and　the　Metropolis-Hastings　(M-H)　sampling　method.　Lastly,　the　designed　method,　i.e.　PSO-DE-MH,　is　verified　by　an　outdoor　experiment　analyses　in　a　section　of　the　South-North　Water　Transfer　Project.　Results　show　that　the　number　of　iterations,　the　average　absolute　error,　the　average　relative　error　and　the　average　standard　deviations　of　the　identification　results　obtained　by　PSO-DE-MH　are　less　than　those　of　PSO-DE　and　DE-MH.　Moreover,　the　relative　error　and　the　sampling　relative　error　of　the　identification　results　under　five　different　measurement　errors　(MEs)　(sigma　=　0.01,　0.05,　0.1,　0.15,　0.2)　are　less　than　9.5%　and　0.2%,　respectively.　The　designed　method　is　effective　even　when　the　standard　deviation　of　the　ME　increases　to　0.2.　Therefore,　the　designed　method　can　effectively　and　accurately　obtain　the　source　information　of　sudden　hazardous　chemical　leakage　accidents.　This　study　provides　a　new　idea　and　method　to　solve　the　difficult　problems　of　emergency　management.