Based on the guaranteed stability approach, a modified adaptive controller is presented for robotic manipulators. The developed control strategy guarantees uniform ultimate boundedness of the tracking errors and parameter errors in the presence of a class of state-dependent uncertainties, and the tracking error can be made arbitrarily small. It is shown that with resettable design parameters and a low-pass-filter form of update equation the performance degradation and destabilisation of robotic systems can be prevented effectively in the practical situation.