In two experiments, the role of spatial ability in learning from an instructional animation versus a series of static pictures was studied. In both experiments, a statistical interaction of spatial ability and type of visualization was obtained: Low-spatial ability students showed poor learning outcome when learning from pictures while high-spatial students did not; when learning from animation, however, learning outcome was independent from spatial ability. The results are in line with an ability-as-compensator hypothesis which states that constructing mental animations from non-dynamic materials needs spatial ability; with animated learning materials, however, spatial ability is not required. No overall differences between static pictures and animation were found.