Behind　the　rapid　development　of　China＇s　logistics　industry,　there　are　problems　of　high　energy　consumption　and　high　pollution.　Under　the　dual　constraints　of　resources　and　environment,　promoting　the　low-carbon　transformation　of　the　logistics　industry　is　the　key　to　achieving　sustainable　development　of　the　logistics　industry.　This　paper　applies　the　epsilon-based　measure　(EBM)　model　which　considers　undesirable　output　and　global　Malmquist-Luenberger　(GML)　index　to　measure　the　logistics　efficiency　under　the　low-carbon　constraints　of　30　provinces　in　China　from　2005　to　2017,　that　is,　the　green　total　factor　productivity　(GTFP),　and　characterizes　its　temporal　and　spatial　evolution　characteristics　through　visualization　and　spatial　analysis　methods.　Then,　this　paper　uses　the　geographically　weighted　regression　(GWR)　model　to　analyze　the　influence　of　industrial　agglomeration　level,　informatization　level,　foreign　direct　investment,　logistics　energy　intensity,　traffic　network　density,　and　technological　innovation　capability　on　the　GTFP　of　the　logistics　industry.　The　findings　of　this　paper　show　that　(1)　during　the　inspection　period,　the　overall　average　GTFP　of　the　logistics　industry　was　0.992,　which　did　not　reach　the　effective　level,　and　the　spatial　differentiation　showed　that　the　average　GTFP　of　eastern　was　greater　than　that　of　in　central,　and　that　of　in　central　was　greater　than　that　of　in　western.　(2)　The　GTFP　of　the　logistics　industry　has　experienced　an　alternating　process　of　rising　and　falling　in　time,　with　large　fluctuations.　Also,　in　terms　of　spatial　dimension,　there　is　a　trend　that　high-level　areas　gradually　gather　to　the　southeast,　and　there　is　significant　spatial　autocorrelation.　(3)　For　the　logistics　industry,　high-efficiency　areas　and　high-output　areas　show　significant　spatial　homogeneity.　(4)　The　estimation　results　of　the　GWR　show　that　the　direction　and　intensity　of　the　multi-dimensional　driving　factors　on　the　GTFP　of　the　logistics　industry　are　different　in　different　regions,　showing　obvious　spatial　non-stationary　characteristics.