Background: There is much support for using cooperative methods, since important instructional aspects, such as elaboration of new information, can easily be realized by methods like "jigsaw." However, the impact of providing students with additional help like a questioning training and potential limitations of the method concerning the (minimum) age of the students have rarely been investigated. Aims: The study investigated the effects of cooperative methods at elementary school level. Three conditions of instruction were compared: jigsaw, jigsaw with a supplementary questioning training and teacher-guided instruction. Sample: Nine third grade classes from three schools with 208 students participated in the study. In each school, all the three instructional conditions were realized in three different classes. Methods: All classes studied three units on geometry and one unit on astronomy using the assigned instructional method. Each learning unit comprised six lessons. For each unit, an achievement test was administered as pre-test, post-test and delayed test. Results: In the math units, no differences between the three conditions could be detected. In the astronomy unit, students benefited more from teacher-guided instruction. Differential analyses revealed that "experts" learned more than students in teacher-guided instruction, whereas "novices" were outperformed by the students in the control classes. Conclusions: Even third graders used the jigsaw method with satisfactory learning results. The modest impact of the questioning training and the low learning gains of the cooperative classes in the astronomy unit as well as high discrepancies between learning outcomes of experts and novices show that explicit instruction of explaining skills in combination with well-structured material are key issues in using the jigsaw method with younger students.