This paper presents a novel adaptive control design procedure for discrete-time nonlinear systems that can be transformed into the parametric-strict-feedback form. This procedure utilizes the tools that are introduced in the companion paper [1], namely, temporal separation of the parameter estimation task from the control task, the implementation of an active identification procedure through a recursive orthogonalized projection estimator, and an input selection algorithm that guarantees complete identification in finite time. However, compared to the output-feedback case presented in [1], the active identification task is now further complicated by the fact that the nonlinearities are allowed to depend on all states, not just the output. This requires a significant modification of the input selection procedure, which utilizes all the measured states to guarantee that the identification task will be completed in less time than in the output-feedback case.