There is considerable current interest in patient satisfaction surveys; among other things they are often used to measure quality in health care. However, there are many unresolved issues concerning them, including how they might best be conducted. Some have challenged their raison d'-etre, arguing that they rely on unproven assumptions -1; Williams, Soc. Sci. Med. 38, 509, 1994]. In this paper we examine some of these assumptions in the light of a recent study of elderly patients' experience of the District Nursing Service. In doing so we argue that users' capacity, or willingness, to evaluate health care services cannot be taken for granted; that patients may have few if any expectations on which to base an evaluation of the care they receive; that the sense of dependency that patients have on service providers can impact on expressions of satisfaction and that, outside the acute sector in particular, we need to take into account the varied nature of the relationships that emerge between patient and service providers. We also argue that elderly patients cannot be treated as consumers unproblematically. While they may be seen as an extreme example, we argue, nevertheless, that many of the issues raised are relevant to much wider populations. This has general implications for research and we suggest that appropriately chosen qualitative methods can provide a useful starting point in studies which aim to establish how, to what extent and whether patients evaluate health services.