On-line　neural　network　(NN)　controllers　are　presented　for　tracking　control　of　rigid　link　electrically　driven　(RLED)　robots　with　uncertainties.　Based　on　backstepping　method　and　Lyapunov　function,　two　categories　of　robust　controllers　with　two-layer　neural　network　identifiers　are　designed　respectively.　Neural　networks　are　used　to　identify　complicated　nonlinearities,　including　modeling　errors,　even　exogenous　disturbances　Owing　to　the　Lyapunov　functions　in　designing　controllers,　stability　of　system　and　uniformly　ultimately　bounded　(UUB)　tracking　errors　can　be　guaranteed.　On-line　weights　tuning　algorithms　are　proposed　Simulations　prove　good　performance　of　the　tracking　control　system.