Honeywell and others have successfully applied dynamic inversion control to many flight control problems. It is a simple and systematic design procedure that guarantees nominal performance. However, unlike H ∞ methods, it does not provide any robustness guarantees. In this paper, we propose a procedure for designing a linear time-invariant dynamic inversion autopilot with guaranteed robustness. We will use the generalized stability margin to choose the dynamics that will be inverted ( A and B matrices), and McFarlane-Glover loop shaping to select the C matrix of the inverted plant, the desired dynamics, and a robust state estimator. The proposed procedure is illustrated by designing a pitch axis autopilot that achieves robust performance with a number of Boeing 747 aircraft models.It is proved in [21] that 0 b(P, C) 1. The larger the value of the generalized stability margin, the more robust the feedback interconnection.Consider the standard McFarlane-Glover loop shaping block diagram depicted in Figure 3. McFarlane-Glover loop shaping has two steps. In the first step, the designer specifies the desiredWe will use C to denote both a controller and the C matrix of a nominal plant. The meaning of C will become apparent from the context.