Positive　matrix　factorization　(PMF)　model　is　one　of　widely　used　technologies　in　pollutant　source　apportionment,　and　its　uncertain　analysis　have　always　been　the　frontier　issue　as　well　as　hotspot.　Three　error　estimation　methods,　including　bootstrap　(BS),　displacement　(DISP)　and　bootstrap　enhanced　by　displacement　(BS-DISP),　were　used　to　evaluate　the　uncertainties　of　source　apportionment　by　PMF　model,　and　heavy　metals　in　soils　in　Dexing,　China　were　carried　out　as　a　case　study.　Six-factor　scenario　was　the　best　solution　for　PMF　model　run,　except　for　Cr　and　Ti,　the　uncertainty　intervals　of　DISP　and　BS　were　between　0.6　and　1.5　times　the　basic　value　of　the　identified　element　and　the　BS-DISP　uncertainty　interval　was　between　0.6　and　1.6　times　the　basic　value　in　the　source　profiles　under　six-factor　scenario,　the　uncertainty　of　the　model　results　was　more　due　to　the　uncertainty　generated　in　the　factor　decomposition　process.　The　three　uncertainty　analysis　methods　could　obtain　the　random　error　and　factor　rotation　error　in　the　operation　of　PMF　model.　Among　them,　the　results　obtained　by　BS-DISP　and　BS　can　assist　in　determining　whether　the　factor　number　was　over-fitting　and　help　understand　the　uncertainty　of　the　source　profile.　While,　DISP　could　be　used　to　understand　the　uncertainty　of　rotation　and　be　used　as　a　method　to　evaluate　the　feasibility　of　rotation　process.　This　study　provides　a　good　sample　for　evaluating　the　reliability　of　soil　heavy　metal　source　apportionment　that　calculated　by　PMF　model.　©　2020,　Editorial　Board　of　China　Environmental　Science.　All　right　reserved.