A　pedestrian　choice　model　of　vertical　walking　facilities　based　on　random　forest　is　established　in　rail　transit　station　considering　four　key　influence　factors:　interlayer　height,　luggage,　the　difference　in　the　number　of　pedestrians　in　front　of　elevator　and　stairway,　and　walking　speed.　This　model　is　verified　with　the　collected　data　of　Changchun　light-rail　transfer　station　and　Beijing　Xi-zhi-men　railway　transfer　station,　China,　and　compared　with　the　choice　model　based　on　support　vector　machine.　Prediction　results　of　these　two　choice　models　are　all　good　and　acceptable.　The　difference　between　the　mean　prediction　accuracies　of　these　two　pedestrian　choice　models　is　small　(0.62%).　Incorporating　the　choice　model　of　vertical　walking　facilities　into　a　cellular　automata　(CA)-based　pedestrian　simulation　model,　and　setting　reminder　sign　in　front　of　pedestrian　choice　zone　to　reduce　conflicts,　pedestrian　choice　behavior　of　vertical　walking　facilities　in　rail　transit　station　is　simulated.　The　simulation　results　indicate　that　efficiency　of　pedestrians　passing　is　improved　with　the　effect　of　reminder　sign,　and　the　distance　between　reminder　sign　and　vertical　walking　facilities　should　be　set　to　be　larger　than　4　m.