In this work we show an application of L1 Adaptive control theory for attitude control of UAVs. We implement the flight control system on a multirotor to show robustness and precise attitude tracking in the presence of modeling uncertainties and environmen- tal disturbances. We choose backstepping control architecture, since the kinematics and dynamics of multirotors in most cases can be written in strict feedback form. We further exploit the fact that the kinematics of the plant, while free of uncertainties, is nonlinear, which makes it highly suitable for dynamic inversion control at each level of backstepping. On the other hand, plant dynamics is uncertain and is affected by environmental distur- bances such as wind gusts, unmodeled dynamics etc. Therefore, we consider 3 variants of the control architecture. The first method uses backstepping to determine the moment demand and augments it with the L1 adaptive controller to account for uncertainties and provide robustness with guaranteed transient performance. The second architecture apply the concept of L1 adaptive backstepping to the same problem; and the third architecture uses L1 backstepping for quaternion representation of the system dynamics, which helps to avoid the singularities associated with Euler angles.