This　study　aims　to　clarify　a　dynamic　optimal　supply　of　parking　permits　considering　the　situation　in　which　reservable　and　unreservable　public　parking　facilities　exist　in　an　area　simultaneously.　We　first　mathematically　established　a　stochastic　optimal　control　model　to　minimize　the　expectation　of　the　total　time　lost.　We　then　analyzed　the　optimality　condition,　i.e.,　Hamilton-Jacobi-Bellman　equation,　by　applying　the　dynamic　programming　principle.　Finally,　with　the　estimated　parameters　in　the　predetermined　optimal　value　function,　we　obtained　the　analytic　solution　of　the　dynamic　optimal　supply　strategy.　The　results　show　that　the　dynamic　optimal　supply　strategy　is　a　feedback　control　strategy,　which　means　the　proposed　strategy　is　a　function　of　queuing　time.　The　dynamic　optimal　supply　strategy　can　be　divided　into　two　patterns　basing　on　the　existence　of　queuing　time.　Taking　the　Guomao　area　in　Beijing　City　as　an　example,　the　Monte-Carlo　experiment　results　show　that　the　dynamic　optimal　supply　strategy　can　save　0.45~1.94　min/veh　compared　with　the　full　supply　strategy　of　parking　permits.　With　the　increase　of　the　uncertainty　in　the　parking　permits　holders＇　behavior,　the　average　time-saving　value　increases,　but　the　coefficient　of　variation　decreases.　Copyright　©　2020　by　Science　Press.