AbstractThis paper generalizes earlier results on adaptive disturbance estimate (innovations) feedback regulation to the case of adaptive tracking, refining some of the techniques and formulations in the process. The adaptations proposed search within the class of all stabilizing two‐degree‐of‐freedom controllers for a nominal plant and minimize a reference signal/disturbance rejection measure. The emphasis in the paper is on issues relevant to the tracking task and simulation results to demonstrate the effectiveness of the approach.The adaptive scheme is based on results concerning the convenient characterization of the class of all stabilizing two‐degree‐of‐freedom controllers in terms of an arbitrary filterQwith a stable proper transfer function matrixQ(s). The fact that the closed‐loop transfer matrices are affine inQ(s) permits a straightforward on‐line least‐squares update of the parameters of the filterQon‐line. The theory for the class of stabilizing controllers is used to set up appropriate signal preprocessing.The direct adaptive schemes proposed turn out to be recursive prediction error schemes which have local convergence properties. When convergent, they ensure enhancement of the performance of a fixed controller in other than the