Conducted a metaanalysis of studies comparing learning outcomes after the use of instructional animations vs static images and identified moderating factors. Relevant studies were identified from searches in the databases SCI, SSCI, ERIC, PsycInfo, and Psyndex as well as from the cross references of identified articles. 26 (out of 57) studies, published between 1973 and 2003, with a total of 3,582 participants were included in the analysis. Overall effect sizes for learning outcome and the influence of the following potential moderator variables were analyzed: representational vs decorational role of animation, type of knowledge to be acquired, video-based vs computer-based animation, level of realism, inclusion of text annotations, inclusion of cues in static pictures, and instructional domain. Results show overall larger (small to medium) effect sizes in learning outcome for animations compared to static images. Learning from animations was even better when animations were representational, that is, they explicitly related to the topic to be learned and were not merely for decoration, and when they were more realistic (e. g. video-based). Finally, it was found that animations were particularly helpful for acquiring procedural motor knowledge and less so for declarative or problem-solving knowledge. It is concluded that noninteractive animations can facilitate learning under specific circumstances.