In two experiments, strategic conflict avoidance maneuvers of 32 licensed pilots were analyzed when they flew a series of missions in a low fidelity (Experiment 1) and a high fidelity (Experiment 2) flight simulator, rendering an outside view and a cockpit display of traffic information. On various legs of the mission, aircraft generating traffic conflicts intruded from above, below and at the same altitude as ownship at various azimuth orientations. Pilot maneuvers were categorized in terms of the dimension of maneuver (vertical, lateral, airspeed), and the direction of change within that maneuver. Decision analysis revealed a strong tendency of pilots to use simple one dimensional maneuvers, and of these, to prefer vertical over lateral and airspeed maneuvers. Airspeed maneuvers in particular were avoided. In Experiment 2, vertical climbing maneuvers were more frequent than descending ones, and vertical maneuvers also tended to be in the opposite direction of the (climbing or descending) traffic. Lateral maneuvers tended to turn away from the approaching traffic. Maneuvers were also categorized in terms of their effectiveness at avoiding a loss of separation. The decision data are interpreted in terms of the extent to which they reflect CDTI data-driven conflict geometry characteristics, versus stereotypes and procedures based upon prior experience