In this paper an improved adaptation law on the upper bound of uncertainties is proposed to guarantee the boundednenss of both states of the plant and the estimated control gains when the boundary layer technique is employed.