This　paper　describes　an　efficient　Motion-Compensated　(MC)　three　dimension　wavelet　video　coding　system.　In　this　system,　MC　temporal　analysis　is　used　to　remove　temporal　redundancy.　Then　an　adaptive　group　of　pictures　structure　according　to　characteristics　of　motion　can　increase　the　coding　efficiency　and　reduce　computational　complexity　and　memory　requirement　at　the　same　time.　Finally,　a　quadtree-based　wavelet　image　coding　method　is　proposed,　which　can　adapt　the　property　of　wavelet　image　coefficient.　The　experimental　results　show　that,　the　coding　system　is　better　than　other　motion-compensated　three　dimension　wavelet　video　coding　methods.