We study the comprehension of a multimedia technical document about gear functioning by young pupils. The research is focused on the effect of three factors on the construction of a mental model: illustration format (animated versus static) signaling cues (presence versus absence) learner-control of information delivery (three rhythms of presentation: speed, slow and self-controlled). The experimental procedure, conducted with 123 children, follows three phases: pre-test, individual passation of the lesson, comprehension test, delayed post-test. The goal of the pre-test is the evaluation of prior knowledge about gears, but also the control of spatial and verbal working memory aptitude and reading performance. The results show an effect of the animated format, of signaling cues and of the rhythm on the immediate comprehension test and delayed test. For the immediate comprehension test, these effects are different according to the kind of comprehension question (recall, transfer, explanation). These effects are maintained at the delayed post-test, for the self-controlled condition and for the pupils with low prior knowledge. The factor information delivery rhythm shows an effect for the delayed post-test. Our observation device of the behaviour of the child during the lesson was specially designed to explore the reading strategies between the medias. (Contains 4 figures, 5 tables and 3 notes.)