In　mobile　cloud　computing　(MCC)　systems,　the　resource　in　both　the　cloud　and　the　mobile　network　should　be　carefully　managed.　Cloud　resource　management　and　radio　resource　management　have　traditionally　been　addressed　separately　in　previous　works.　In　this　paper,　we　propose　to　jointly　study　dynamic　cloud　and　radio　resource　management　so　as　to　improve　end-to-end　performance　of　adaptive　video　transmissions　in　MCC　systems.　Video　application　quality　of　service　performance,　distortion,　is　adopted　as　the　performance　measure.　An　important　video　application　layer　parameter,　intra-refreshing　rate,　is　optimized　to　improve　the　video　distortion　performance.　We　formulate　the　problem　as　a　stochastic　restless　bandits　optimization　problem,　which　facilitates　the　distributed　MCC　architecture　and　simplifies　the　computation　and　implementation　due　to　its　＂indexibility＂　property.　Simulation　results　are　presented　to　show　the　effectivenes　of　the　proposed　scheme.