Aimed　at　the　problem　that　storage　and　retrieval　in　the　underground　3D　garage　took　users　a　long　time,　the　number　of　car　elevators　in　the　cylindrical　underground　3D　garage　was　investigated.　According　to　the　traffic　survey　data　for　a　design　example,　the　storage-retrieval　process　was　simulated　by　MATLAB　program.　The　probability　distribution　of　arrival　time　of　users　and　running　time　of　the　elevators　was　analyzed　and　obtained　by　probability　theory　and　mathematical　statistics　respectively.　Based　on　the　queuing　theory,　the　influences　of　car　elevator　number　on　storage-retrieval　time　were　investigated　by　the　multi-server　queuing　model　with　finite　system　capacity.　And　then　combined　with　the　car　storage-retrieval　time　which　users　could　bear,　the　number　of　car　elevators　could　be　determined　with　analyzing　the　influence　of　the　probability　distribution　parameter　λ　on　storage-retrieval　time.　The　results　show　that:　the　probability　distribution　of　arrival　time　of　users　follows　a　Possion　distribution　with　parameter　λ　being　29.71　s·veh-1.　The　probability　distribution　of　running　time　of　1,　2　and　3　elevators　follows　a　Possion　distribution　with　the　parameter　λ　being　51.82,　26.67,　18.66　s·veh-1　respectively.　And　the　mean　sojourn　time　of　storage-retrieval　is　276.6　s　and　75.8　s　respectively　when　2　or　3　elevators　are　set　in　the　10-layered　underground　garage.　And　it　is　suggested　that　3　car　elevators　should　be　equipped　in　the　10　layers　of　underground　garage.　Combined　with　influence　of　the　probability　distribution　parameter　λ　of　arrival　time　of　users　on　car　storage-retrieval　time,　2　car　elevators　should　be　equipped　in　3　to　8-layered　underground　garage　and　3　car　elevators　should　be　set　in　9　and　more　than　9-layered　underground　garage.　The　established　method　will　provide　a　theoretical　basis　for　the　calculation　on　the　number　of　car　elevators　to　be　equipped　for　this　kind　of　garages.　©　2017,　Editorial　Department　of　China　Journal　of　Highway　and　Transport.　All　right　reserved.