This article compares the academic performance and classroom behaviour of pupils when taught new concepts or engaged in problem solving in sessions organised either as cooperative group work or whole class, teacher directed instruction. Comparisons of attainment were made in classes of pupils aged 11 to 14 years (Key Stage 3) in English, mathematics and science. Pupils were also observed, mainly during the introductory phase of the topic under investigation, using a specially designed structured observation schedule. The attainment results suggest that a grouping approach is as effective, and in some cases more effective, than when whole class teaching is used. Classroom observation indicated that there were more sustained, higher cognitive level interactions when pupils worked in groups than during whole class discussions. It is argued in conclusion that the group work results could be improved still further if teachers gave more attention to training pupils to work in groups and if more time was given to debriefing after group work. (Contains 5 tables, 4 figures and 2 notes.)