Muscle　fatigue,　as　a　common　physiological　phenomenon,　has　attracted　much　attention　in　the　fields　of　rehabilitation　and　athletic　training.　A　wearable　technology　for　monitoring　the　muscle　fatigue　anytime　and　anywhere　is　urgently　needed.　In　this　paper　we　apply　Electrical　impedance　myography　(EIM)　technique,　usually　used　for　non-invasive　detection　of　neuromuscular　diseases　with　the　four-electrode　array,　for　evaluation　of　the　local　muscle　fatigue　status　via　the　variation　of　electrical　impedance.　An　equivalent　multilayer　inhomogeneous　3D　finite　element　model　of　human　arm　was　built　in　order　to　optimize　the　four-electrode　configuration　to　improve　EIM　detection　sensitivity.　Current　density　in　muscle　layer　and　differential　potential　of　induction　electrodes　were　selected　as　the　evaluation　indexes　for　optimization.　Then　the　in　vivo　experiments　of　dynamic　contraction　with　different　maximal　voluntary　contractions　(MVC)　were　performed　on　the　biceps　brachii　muscle　of　eight　healthy　volunteers.　The　results　showed　that　muscle　resistance　(R)　decreased　almost　8　Omega　from　the　completely　relaxed　muscle　to　exhaustion,　which　is　the　same　trend　as　for　the　median　frequency　(MF)　of　measured　surface　electromyography　(sEMG)　signals.　The　model　and　experiments　in　this　paper　indicate　the　feasibility　and　efficiency　of　EIM　for　detection　of　muscle　fatigue　using　wearable　devices.