Introduction/Objective: To develop an improved theoretical understanding, derived from the experience of a city-wide implementation project, about how best to implement evidence.

METHODS: This paper reports the model developed by a large, on-going and successful implementation project to explain its successes and examine its failures. Quantitative and qualitative data underpin this theory which was developed via an iterative process with the independent evaluator of the project.

Results: In its work to increase the use of aspirin in secondary prevention the project recruited more than 65 practices in one large conurbation. Typically prescriptions for aspirin to high risk groups rose from around 50% to 80% and as a result of involvement with the project >2.500 patients at high risk of CHD and stroke received aspirin for the first time. 'Ownership' 'opinion leaders' 'guidelines' were all found to be less important than the health services literature imply. More useful concepts were found to include: the active translation between the different world views to be found in any health service: techniques to build coalitions between key individuals supporting the proposed change; using marketing and selling techniques; reassessing the role of the rationalist approach to changing clinical practice.

Discussion: Current models of development and implementation are structured around a number of assumptions, which are rarely questioned. Understanding how these assumptions limit thinking, together with developing other models more directly based in practice will help to improve our ability to generate change in clinical behaviour.