Evaluation of large-scale professional development programmes in a cost-effective but detailed manner presents challenges. Much of the research evidence on the effectiveness of professional development in science education has been the result of small-scale studies that use clustered or stratified sampling methods. Using the evidence collected from the national Network of 10 Science Learning Centres in the UK, this research examines the nature and extent of change on a national scale from large numbers of teachers' involvement in professional development courses. The network administers a common system of "embedded evaluation", which integrates completion of evaluation forms in the training process, thus making a cross-network study of impact possible. The research employed a triangulated approach of analysis of reported impact from different sources including impact toolkit forms, interviews with participants, a survey of their line managers and evidence from course leaders. The research shows that besides the direct impact on teachers' own classroom practice, there is evidence of wider change involving other teachers and on whole school practice. Use of an evaluation toolkit embedded in the professional development course is regarded as a key process in supporting change and monitoring progress. (Contains 3 figures and 4 tables.)