Abstract: Patient centred approaches to care are increasingly advocated by consumers, clinicians and policy makers, and patient centred communication skills are being incorporated into the training of health care providers. However, patient centred care is difficult to define and relatively little empirical work has been undertaken to assess the effects of interventions to promote it within clinical consultations. This paper discusses the conceptual and other methodological difficulties that have arisen in the course of undertaking a systematic review of the effects of interventions to promote patient centred approaches in clinical consultations. These include: * The initial scoping of the review, and how this was influenced by the fact that you cannot simply randomise people to receive patient centred care but only to interventions that aim to promote or deliver it. * The difficulties of defining patient centred care and specifying appropriate inclusion criteria in the context of a literature that includes diverse definitions and applications of patient centredness. * The difficulties of applying inclusion criteria at the boundaries. For example, we grappled with the question of what constitutes 'sufficient' patient centredness for our purposes. * The fact that there is no valid and reliable measure of patient centredness that could be used either as an outcome measure or to check that the intervention was working as intended. * Attempts to develop meaningful ways of grouping diverse interventions, for example by assessing the intensity of interventions (in terms of patient centredness and of the training strategies used) or the types of consultation settings in which they were delivered. In this poster we will explain how we dealt with these issues. We will comment on the advantages of having a multidisciplinary review team with experiences of different health care systems. We will also suggest that for reviews in which conceptual meanings are contested, protocols may need to be developed in a more iterative fashion than tends to be assumed within the Collaboration and protocol development manuals.