Twenty-four nonpilot, volunteer subjects were tested on three flight tasks while flying four basic aircraft attitude presentations (moving horizon, moving airplane, frequency-separated, and kinalog) in a light twin-engine aircraft simulator providing three types of motion cues (no motion, standard GAT-2 motion, and washout motion). The flight tasks involved conflicting visual and vestibular cues and included disturbed attitude tracking, command flight path tracking in both pursuit and compensatory modes, and a series of recovery trials from discrete unknown attitudes. To provide a basis for comparison, the present simulator study closely replicated the procedures used in the Roscoe and Williges (1975) flight experiment. The frequency-separated display yielded performances at least equivalent and in some cases superior to those obtained with the conventional moving horizon display. Either type of simulator motion resulted in better disturbed attitude tracking performance than no motion, and washout motion provided stereotypic control responses in recovery from unknown attitudes most closely corresponding to those obtained in flight. It was concluded that care must be used in generalizing simulator results to flight performance when no physical motion cues or inappropriate ones are present in the simulator.