Eye movements were investigated amongst participants who preferentially use an egocentric or an allocentric frame of reference during navigation through computer simulated tunnels. Performance was highly accurate even though the tunnel passages contained only sparse visual flow and no differences in homing accuracy between subjects using one or the other reference frame was observed. Analyses of eye-movements revealed that gaze was centered on the tunnel's visual centroid during straight segments. However, during turns mean gaze position was directed toward the outer wall. As the angle of turn increased, the prevalence of overall eye movements and the laterality of gaze were greater than during turns of lesser angle. Even though the strategy groups reacted based on distinct reference frames, comparable patterns of eye movements were revealed for both strategies. The data describe how information during navigation through sparse visual environments is selected and demonstrate that the preferential use of an egocentric or an allocentric frame of reference is independent of eye-movement patterns. Thus, a purely cognitive basis for individual differences in reference frame usage can be assumed.