In this paper, the adaptive augmentation of the attitude control system for a multirotor unmanned aerial vehicle is considered. The proposed approach allows to combine a baseline controller with an adaptive one and to disable or enable the adaptive controller when needed, in order to take the advantages of both the controllers. To improve transient performance with respect to the standard model reference adaptive controller, an observed-based approach is exploited. The adaptation law is based on the error between the output of an observer of the nominal closed-loop dynamics and the actual output of the system with uncertainties. Experimental results obtained by testing the proposed approach on a quadrotor unmanned aerial vehicle are presented to compare the performance, in terms of disturbance rejection, with respect to the baseline controller and to a [Formula: see text] adaptive augmentation scheme.