Two theories of the spacing effect are compared in the context of aircraft recognition: working memory reconstruction and encoding variability. Participants (130 men and 110 women) viewed enhanced silhouettes of aircraft in one of four presentation schedules: massed once, massed repeated, distributed fixed, and distributed random. The two distributed schedules were designed to differ only on structural components. Thus the encoding variability theory predicts that the distributed-random schedule will be superior to the distributed-fixed schedule, whereas the working memory reconstruction theory predicts no difference between the two. Measures of recognition discriminability on novel view stimuli indicated that men, but not women, showed a recognition with the distributed-random schedule superior to that with all other schedules. Thus the men's data were consistent with the encoding variability hypothesis. The gender difference suggests the importance of processing style, familiarity, and motivation in aircraft recognition. These results have implications for aircraft recognition training and instructional design.