Nowadays,　wireless　communications　are　entering　the　5G　era,　in　which　Mobile　Edge　Computing　(MEC)　and　Ultra　Dense　Network　(UDN)　architecture　are　the　two　core　technologies.　In　UDN　architecture,　deploying　MEC　servers　at　both　macro　eNodeB　and　small　eNodeB　can　effectively　reduce　delay　and　energy　consumption,　and　improve　the　user＇s　Quality　of　Experience　(QoE).　However,　the　access　selection　is　challenging　due　to　the　dense　allocation　of　small　eNodeBs.　In　this　paper,　we　propose　a　user　access　selection　algorithm　considering　MEC　in　UDN.　A　QoE-aware　cost　function　related　to　energy　consumption　and　service　delay　is　adopted　as　the　performance　metric.　The　energy　consumption　is　calculated　as　the　uplink　transmission　power　of　User　Equipment　(UE),　and　the　service　delay　is　counted　as　the　sum　of　the　transmission　delay　from　UE　to　the　MEC　server　and　the　processing　delay　of　user　services　at　the　MEC　server.　This　method　is　compared　with　the　traditional　optimal　channel　quality　access　method.　Numerical　results　show　that　the　proposed　method　can　effectively　reduce　the　delay　and　energy　consumption,　and　thus,　improve　the　QoE　of　users.　©　2017　IEEE.