The　surface　electromyography　(sEMG)　signal　is　the　sum　of　the　action　potentials　generated　by　the　active　motor　units　and　detected　over　the　skin,　which　has　great　performance　on　the　recognition　of　human　movements　and　the　diagnosis　of　injuries　because　of　its　strong　muscle　specificity　and　differences　in　exercise　patterns.　This　paper　employs　the　Cerebellar　model　neural　network　(CMNN)　as　a　classifier　of　the　ankle　motion　based　on　sEMG.　We　testify　our　method　on　the　data　recorded　from　six　healthy　subjects　who　are　on　isokinetic　ankle　eversion　and　inversion.　The　results　show　that　the　classification　accuracy　is　higher　than　96.9%　with　less　training　times.　For　the　future　application,　the　CMNNs　can　be　employed　to　predict　the　ankle　motions　in　real-time　to　control　exoskeleton　robot.