As　network　mobile　traffic　grows　rapidly,　how　to　allocate　various　network　resources　to　improve　the　efficiency　of　content　distribution　and　quality　of　experience　is　an　urgent　problem　to　be　solved　for　current　delay-sensitive　services.　To　efficiently　reduce　network　delay,　the　idea　of　cloud-edge　cooperation　has　recently　drawn　great　attention　and　been　explored.　However,　the　joint　allocation　of　caching,　computing　and　transmission　resources　has　been　lack　of　in-depth　study,　and　content　popularity　closely　related　to　traffic　distribution　has　been　largely　ignored　in　existing　works.　In　this　paper,　we　propose　a　Q-learning　based　delay-aware　content　delivery　scheme　in　a　three-layer　cloud-edge　cooperation　network.　In　our　solution,　the　queueing　theory　is　used　to　formulate　the　content　distribution　process　as　a　delay　minimization　problem,　where　a　Q-learning　policy　is　utilized　to　make　optimal　cache　placement　decisions　to　improve　network　delivery　and　delay.　Simulation　results　demonstrate　that　our　proposed　model　performs　much　better　than　the　related　popular　solutions　and　close　to　its　centralized　alternative.　©　2021　IEEE.