The　skip-stop　operating　mode　for　urban　rail　transit　systems　defines　three　types　of　stations:　A,　B,　and　AB　stations.　In　this　strategy,　all　the　trains　stop　at　AB　stations,　and　A　(B)　trains　stop　at　only　A　(B)　stations.　Skipping　some　stops　can　increase　the　operating　speed,　which　leads　to　direct　benefits　by　reducing　the　in-vehicle　time　for　passengers　and　improving　the　operational　efficiency　for　operators.　With　the　skip-stop　mode,　passengers　have　to　transfer　at　an　AB　station　when　the　trains　they　take　do　not　stop　at　their　destination,　and　the　extra　transfer　waiting　time　depends　on　the　AB　station　distribution　and　the　corresponding　headways.　In　this　paper,　we　develop　a　train　scheduling　optimization　model　to　minimize　the　total　passenger　travel　time　so　that　passengers　can　get　to　their　destinations　faster　without　experiencing　excessive　transfer　waiting　times.　A　two-phase　approach　based　on　a　genetic　algorithm　is　developed　to　solve　this　mixed-integer　nonlinear　problem.　Finally,　the　proposed　train　scheduling　optimization　model　is　applied　to　an　idealized　corridor　and　a　real-world　scenario　based　on　Line　6　of　the　Beijing　URT　network.　An　analysis　of　the　type　of　scenarios　in　which　skip-stop　operations　are　applicable　and　the　efficiency　of　different　types　of　trips　is　presented.　The　results　show　that　a　short　headway　(below　5-6　min)　is　favorable　for　a　skip-stop　operation　and　there　needs　to　be　at　least　one　AB　station　for　everyfive　stations.　The　average　travel　time　per　passenger　can　be　reduced　by　more　than　1　min　for　the　Beijing　scenario　under　the　optimized　skip-stop　operation.