In educational settings, computers provide specific possibilities of visualizing information for instructional purposes. Besides the use of static pictures, computers can present animated pictures which allow exploratory manipulation by the learner and display the dynamic behavior of a system. This paper develops a theoretical framework for analyzing the comprehension of static and animated pictures and for selecting adequate kinds of visualization for instructional purposes. An experiment was performed in which different groups of university students (n=80) had to understand a complex subject matter with the help of computer-based hypertext that contained either static or animated pictures. Results indicate that the structure mapping process is influenced both by the semantic content to be conveyed and by surface characteristics of the picture. The visualization effect is more pronounced with static pictures than with animated pictures. Generally, learners do not use picture information instead of text information and vice versa; instead, text comprehension and picture comprehension stimulate each other. Since animated pictures can present different states of a subject matter, they provide more information to a learner and require more processing than static pictures. Animated pictures reduce the cognitive load of knowledge acquisition, as they provide external support for mental simulations. However, animated pictures may result in a less intensive processing, because they prevent individuals from performing relevant cognitive processes by themselves. Since learners merely have to follow an externally displayed simulation, they may process the subject matter more superficially than they would do with static pictures. (Contains 52 references.) (Author/SWC)