The Scottish Executive (SE) commissioned the first phase of a National Coronary Heart Disease (CHD) Demonstration Project, Have a Heart Paisley (HaHP), in 2000. HaHP was a complex community-based partnership intervention. An independent evaluation of HaHP (phase one) was commissioned by the SE in 2001. This thesis presents the learning from the evaluation. The first aim is to identify the key implementation, evaluation and policy lessons to result from the evaluation. The second is to contribute to learning about how best to evaluate complex community-based interventions. The evaluation consisted of four approaches: a theory-based approach (the Theories of Change); the mapping of the context; a quasi-experimental survey; and, a range of integrated case studies. This thesis uses the programme logic (the intervention’s Theories of Change) articulated by the HaHP stakeholders to integrate the results from each of the evaluation approaches. HaHP (phase one) did not achieve significant changes in population level CHD risk factors, behaviours, morbidity or mortality. Like many previous community-based CHD interventions HaHP did not fully implement its intended Theories of Change. HaHP’s activities were not consistently based on best practice. It did not articulate or implement clear strategies for addressing health inequalities. The project delivered mainly individually focussed, ‘downstream’ interventions and struggled to achieve wide-scale local service, policy and agenda changes. It did, however, make progress with regard to improving partnerships and jointly delivering interventions. The findings from HaHP add to existing evidence that large-scale behaviour and cultural change will only be achieved through national action and the increasing use of ‘upstream’, legislative, or policy solutions, or changes in mainstream services and organisations. Activity in localised demonstration projects can add to such change rather than create it. The Theories of Change approach claims to improve planning and implementation, enhance evaluation, and address attribution. The approach (as applied within this evaluation) provided substantial amounts of formative feedback that was of use for improving programme implementation. This learning, however, was not always acted upon.