The growing prevalence of heart failure and the cost associated with its management has become a major burden for most health care systems worldwide.We review the evidence from randomized studies of innovative models of care delivery for patients with heart failure, refer to some of the most influential non-randomized studies, and discuss the implications of the available evidence for practitioners, policy makers and researchers. Acknowledging that the relevance of evidence depends on the user's needs, we conclude that the likely impact of most models of care on health outcomes and resource utilization is likely to be modest. New approaches for design and evaluation are therefore required. Given the dynamic complexity of the health service environments in which any such models of care must be implemented, the future development of innovative models of care delivery would benefit from closer collaboration between service users, providers, policy makers and multi-disciplinary researchers, as well as more rigorous evaluation.