Forward　osmosis　(FO)　has　gained　considerable　interest　as　an　attractive　membrane-based　desalination　technology　due　to　its　reversible　fouling　and　potentially　low　costs.　Thin-film　composite　(TFC)　membranes,　composed　of　a　microporous　substrate　and　a　selective　polyamide　layer,　have　been　widely　used　for　FO.　Compared　with　the　substrate　in　pressure-driven　TFC　membranes　(e.g.,　reverse　osmosis　and　nanofiltration　membranes),　the　substrate　in　TFC　FO　membranes　has　a　more　significant　impact　on　membrane　performance　due　to　its　extra　effect　on　the　membrane　structural　parameter　that　determines　internal　concentration　polarization　(ICP).　Many　researchers　have　not　differentiated　the　substrate　surface　properties　from　the　substrate　bulk　properties,　resulting　in　misleading　conclusions.　This　review　clarifies　the　individual　roles　of　the　substrate　surface　and　the　substrate　bulk　in　TFC　FO　membrane　performance.　We　analyze　how　the　substrate　surface　properties,　such　as　the　surface　hydrophilicity,　porosity,　pore　size,　pore　size　distribution　and　pore　number　density,　impact　the　PA　layer　formation　during　IP,　thereby　influencing　the　PA　layer　permeability/selectivity　and　thus　FO　performance.　We　also　evaluate　how　the　substrate　bulk　properties,　such　as　the　bulk　porosity,　tortuosity　and　pore　structure,　affect　the　membrane　structural　parameter,　thereby　influencing　ICP　and　thus　FO　membrane　performance.　A　desirable　substrate　surface　should　have　reasonably　high　hydrophilicity　(e.g.,　water　contact　angles　of　40°–60°),　high　porosity,　small　pore　size,　narrow　pore　size　distribution　and　high　pore　number　density.　A　desirable　substrate　bulk　should　have　high　porosity,　low　tortuosity　and　interconnected　pore　structures,　thereby　minimizing　the　membrane　structural　parameter　and　thus　ICP.　This　review　highlights　the　significant　roles　of　the　substrate　surface　and　bulk　properties　in　FO　membrane　performance.　It　provides　significant　insights　into　developing　all　kinds　of　high-performance　TFC　membranes　(e.g.,　FO,　reverse　osmosis　and　nanofiltration)　by　tailoring　the　substrate　properties.　©　2022　Elsevier　B.V.