A new least-squares estimator with nonlinear data weighting is developed and used to adaptively control a simple discrete-time nonlinear system. A global stability result is obtained through Lyapunov analysis without assuming any growth conditions on the nonlinearities.