New technologies enable flexible combinations of text and interactive or non-interactive pictures. The aim of the present study was to investigate (a) whether adding pictures to texts is generally beneficial for learning or whether it can also have detrimental effects, (b) how interactivity of pictures affects learning, (c) whether the visualization format of pictures affects the structure of the learner's mental model, and (d) whether the visualization format modifies the effects of interactivity. One hundred university students were randomly assigned to five groups. In four groups, a text about the different daytimes and days on the earth was combined with interactive or non-interactive pictures of different visualization formats. In the fifth group, the text was presented without pictures. According to the results, adding pictures to text was neither beneficial nor harmful for learning. In terms of learning efficiency, however, learning from "text only" was "more" successful than learning from text and pictures. Interactivity was beneficial for one learning task, but not for the other task. The visualization format affected participants' interaction with pictures, but not the learning outcomes; however this effect was not influenced by interactivity. Implications for multimedia design and for further research are pointed out. (Contains 6 figures and 1 table.)