The rationale of evidence-based decision-making is to inform the decision-making process with information relevant to the decisions being taken. In this paper the models of research and analytical approaches used to generate the evidence are shown to be generally not 'decision-informing'. The researcher's interest in health care interventions has led to the development and use of designs which strip the research of contextual issues and hence represent a major departure from both the underlying notions of the complex pathways to health and the empirical findings concerning the importance of population context. In this way, the evidence-based approach, dominated by a focus on health outcomes from health care interventions, overlooks the notion that society is not a 'level playing field'. Decisions based on research 'evidence' of this type risk redeploying resources inefficiently and in ways which systematically favour those groups with favourable 'prospects for health' (or non-health care determinants of health), and the conditions that those groups in society tend to suffer from, and away from those groups with less favourable prospects for health. Existing approaches to informing the decision-making process could be enhanced by broadening the scope of the research to incorporate relevant determinants of health in both the specification of the problem and the selection of methods of analysis that enable us to explore the complex pathways to health.