The use of cooperative learning in secondary school is reported--an area of considerable concern given attempts to make secondary schools more interactive and gain higher recruitment to university science courses. In this study the intervention group was 259 pupils aged 12-14 years in nine secondary schools, taught by 12 self-selected teachers. Comparison pupils came from both intervention and comparison schools (n = 385). Intervention teachers attended three continuing professional development days, in which they received information, engaged with resource packs and involved themselves in cooperative learning. Measures included both general and specific tests of science, attitudes to science, sociometry, self-esteem, attitudes to cooperative learning and transferable skills (all for pupils) and observation of implementation fidelity. There were increases during cooperative learning in pupil formulation of propositions, explanations and disagreements. Intervened pupils gained in attainment, but comparison pupils gained even more. Pupils who had experienced cooperative learning in primary school had higher pre-test scores in secondary education irrespective of being in the intervention or comparison group. On sociometry, comparison pupils showed greater affiliation to science work groups for work, but intervention pupils greater affiliation to these groups at break and out of school. Other measures were not significant. The results are discussed in relation to practice and policy implications. (Contains 4 tables.)