This meta-analytical review focuses on the role of spatial ability when learning with pictorial visualizations. By means of selective theoretical review and meta-analysis (the latter regarding 27 different experiments from 19 studies), several sub-factors of spatial ability are considered as well as dynamic and non-dynamic, interactive and non-interactive visualizations. An overall effect of r = 0.34 (95%-CI 0.28 to 0.39) demonstrating a medium advantage for high-spatial-ability learners when working with visualizations is calculated. More importantly, two moderators could be identified: Learners with low spatial ability can be significantly supported by a dynamic instead of a non-dynamic visualization as well as by 3d- instead of 2d-illustrations. Results are discussed in consideration of contemporary theories of multimedia learning.