Arguments for the use of interactive three-dimensional (3D) simulations in education and training depend to a large extent on an implicit assumption that a more accurate and complete spatial cognitive model can be formed through active user-controlled exploration of such an environment than from viewing an equivalent animation. There is a similar implicit assumption that the viewing of animated view changes provides advantages over the viewing of static images due to the value of optical flow. The results to date, however, do not clearly support these assumptions. In particular, the findings of Peruch, Vercher and Gauthier (1995) and Christou and Bulthoff (1999) conflict in relation to the importance of active exploration and of optical flow. This paper reports the results of two studies exploring the importance of active exploration and of optical flow for spatial learning in 3D environments. The results indicate that active exploration can provide greater spatial learning than viewing of animations, but only if there is an alignment between the task goals during this exploration and the spatial learning being tested. Additionally, the results suggest that a set of well-chosen static views of the environment can in some cases allow the formation of as complete a spatial cognitive model as a set of animated views. The paper concludes with an analysis of the methodologies used by Peruch et al. and by Christou and Bulthoff in light of the findings reported here, leading to a new explanation for their conflicting results.