Ammonia　(NH3)　has　been　considered　to　be　a　promising　hydrogen　storage　medium　owing　to　the　carbon-free　features,　easy　liquefaction　storage,　low　transportation　costs,　and　potential　ammonia　production　from　renewable　energy　sources.　On-site　hydrogen　production　using　ammonia　decomposition　offers　a　sustainable　and　cost-efficient　solution　for　hydrogen　refuelling　stations,　and　separating　H-2　from　a　H-2-N-2　mixture　is　a　necessary　step　to　obtain　high-purity　H-2　(＞99.97%).　In　the　scale　of　a　hydrogen　refuelling　station　(similar　to　300　Nm(3)　h(-1)),　using　pressure-swing　adsorption　(PSA)　is　not　feasible　owing　to　its　low　recovery,　while　using　polymeric　membranes　cannot　meet　the　H-2　purity　demand.　To　achieve　both　a　high　H-2　purity　and　high　H-2　recovery,　we　developed　a　physical-chemical　system　model　of　a　300　Nm(3)　h(-1)on-site　NH3-fed　hydrogen　refuelling　station　to　optimize　a　H-2　purification　subsystem,　and　furthermore　predicted　the　system　efficiency　and　economic　feasibility.　We　validated　our　system　model　using　experimental　data,　and　compared　eight　different　scenarios　of　H-2　purification　subsystems.　The　results　reveal　that　a　NH3-fed　on-site　hydrogen　refuelling　station　using　a　＂PSA-to-membrane＂　subsystem　is　a　feasible　method　of　producing　high-purity　H-2　with　a　H-2　recovery　greater　than　95%,　which　is　29%　higher　than　a　system　only　using　PSA.　Correspondingly,　the　system　efficiency　increased　from　59.1%　to　85.37%,　and　the　total　specific　cost　was　reduced　by　22%　to　4.31　euro　per　kg.　The　feedstock　cost　accounts　for　74%　of　the　total　specific　cost.　Using　our　optimized　hybrid　H-2　purification　subsystem,　the　H-2　production　cost　of　the　NH3-fed　on-site　hydrogen　refuelling　station　was　at　least　15%　lower　than　other　carbon-free　routes　(such　as　electrolysis,　solar　thermolysis,　photo-electrolysis,　etc.),　and　comparable　to　that　of　a　methane　steam　reforming　system　with　carbon　capture　and　storage.