Dynamic inversion is presented as a general tool for designing decoupling control laws for multivariable systems, with the difference that instead of using conventional, static, full-state feedback, a dynamic compensator is used. This has the advantages that fewer measurements are required and that the designer has better control over the bandwidth of inner feedback loops. This feature is important because it allows the control law to be made less sensitive to the effects of unmodeled sensor and actuator dynamics, which can be destabilizing. The main focus is on applications to linear systems, although some results may extend to nonlinear systems. The issue of zero dynamics and nonminimum phase is discussed in the linear context. Three examples are presented relating to decoupling of lateral-directional dynamics for highly maneuverable aircraft.