The purpose of this study was to assess the effectiveness of two generative learning strategies, concept maps and manipulation of objects, and to determine if either works better with individual learners or in cooperative learning groups. A total of 80 sixth-grade students in science education were randomly assigned to group or individual conditions and to one of the two experimental treatments. Experimental treatments were changed between first and second post-test. Long-term retention was evaluated with a third and delayed post-test. Students starting with concept maps showed higher achievement on delayed post-test than students beginning with manipulation of objects. No difference was found between students working as individuals or cooperative student teams. Furthermore, a significant interaction between generative learning strategy and grouping condition was revealed. Implications for sequencing generative learning strategies are discussed. (Author)