The performance, accuracy, and observability of a model-free angle of attack and angle of sideslip estimator are presented. The estimator does not require an aircraft dynamic model; rather, it only relies on measurements from a GPS receiver, an inertial measurement unit, and a pitot tube. The estimator is an inertial navigation system (INS)/GPS extended Kalman filter augmented with the states to account for wind and an additional measurement from a pitot tube. It is shown that the estimator is conditionally observable. Conditions (maneuvers) that enhance its observability are identified. A bound on the angle of attack and angle of sideslip estimate uncertainties is derived. The effect of INS/ GPS, horizontal and vertical wind uncertainty on the accuracy of angle of attack and angle of sideslip estimate is assessed. Simulation and flight-test results of the method are presented. The results show that the 1 − σ bound on a small, slow-flying unmanned aerial vehicle for angle of attack and sideslip angle estimates are about 5 and 3 deg, respectively.