Landing is a challenging flight phase for automatic control of fixed-wing aircraft. For unmanned air vehicles in particular, it is imperative that model uncertainty be considered in the control synthesis. These vehicles tend to have limited sensors and instrumentation yet must achieve sufficient performance in the presence of modeling uncertainties and exogenous inputs such as turbulence. Quantitative feedback theory has been reported in the literature for design of automatic landing control laws, but none of these controllers has been flight-tested. In this paper, quantitative feedback theory is employed to synthesize robust discrete-time controllers for automatic landing of an unmanned air vehicle. A low-cost flight vehicle with standard aileron, rudder, elevator, and throttle controls is used. Dynamic simulation is conducted using uncertain aircraft models and sensor noise profiles derived from flight hardware. Controllers are initially synthesized in deterministic simulations. Control validation is performed using a Monte Carlo analysis of stochastic simulations. Sources of uncertainty considered are sensor noise, model uncertainty, and static winds. Landing-phase simulations presented in this paper indicate a routinely high probability of a successful landing in relatively calm wind conditions. The flight-testing process is discussed, and time histories from two automatic landings are presented. Dynamic responses in flight test are found to be similar to the simulation, but a significant amount of control redesign is still required to achieve adequate experimental performance. The methodology is judged to be a promising candidate for an automatic landing controller for unmanned air vehicles.