Utilizing　rail　transit　system　for　collaborative　passenger-and-freight　transport　is　a　sustainable　option　to　conquer　urban　congestion.　This　study　pro-poses　effective　modeling　and　optimization　techniques　for　planning　a　city-wide　metro-based　underground　logistics　system　(M-ULS)　network.　Firstly,　a　novel　metro　prototype　integrating　retrofitted　underground　stations　and　newly-built　capsule　pipelines　is　designed　to　support　automated　inbound　delivery　from　ur-ban　logistics　gateways　to　in-city　destinations.　Based　on　four　indicators　(i.e.　unity　of　freight　flows,　regional　accessibility,　environmental　cost-saving,　and　or-der　priority),　an　entropy-based　fuzzy　TOPSIS　evaluation　model　is　proposed　to　select　appropriate　origin-destination　flows　for　underground　freight　trans-port.　Then,　a　mixed　integer　programming　model,　with　a　well-matched　solu-tion　framework　combining　multi-objective　PSO　algorithm　and　A*　algorithm,　are　developed　to　optimize　the　location-allocation-routing　(LAR)　decisions　of　M-ULS　network.　Finally,　real-world　simulation　based　on　Nanjing　metro　case　is　conducted　for　validation.　The　best　facility　configurations　and　flow　assign-ments　of　the　three-tier　M-ULS　network　are　reported　in　details.　Results　confirm　that　the　proposed　algorithm　has　good　ability　in　providing　high-quality　Pareto-optimal　LAR　decisions.　Moreover,　the　Nanjing　M-ULS　project　shows　strong　economic　feasibility　while　bringing　millions　of　Yuan　of　annual　external　benefit　to　the　society　and　environment.