Objective: To determine any social class differences in place of death of cancer patients in South Bristol; to explore the experience of carers; and to identify inequalities in access to palliative care., Design: Two-part study: (1) A cross-sectional survey of all 960 cancer deaths in South Bristol between September 1999 and December 2002. (2) A qualitative in-depth interview study of 18 carers of patients who died of cancer during the same period in South Bristol. Fourteen of those who died were from social class IIIM (manual), IV and V (i.e. lower social classes)., Main Outcomes: Place of death of patients according to social class and geographical distance from the hospice. Carers' accounts of the way in which illness and death were conducted, and their response to the management of death and dying., Results: The cross sectional survey showed that patients from social class V were less likely to die in the hospice. This finding was independent of geographical proximity. In the qualitative study, no class specific beliefs about death and dying were identified. Attitudes to the way dying should be conducted were common across the classes. Families expected to be present and centre stage at the time of death and for it to be conducted in a dignified and personal manner. Health care staff in all settings supported them in this aim. No one in this study died without a family member present. Some elderly carers were less open than younger carers in the way they talked about death and did not wish to be present at the death. Some carers from social class IV and V were less active in seeking information or asking for hospice admission than carers from other classes. Unrealistic expectations about the availability of hospice beds were common to all carers. Anxiety was common among carers. It was reduced by the provision of reliable and consistent healthcare support, by information provided in a timely and sensitive way, and by open and shared decision making between carer and patient. Most important in reducing anxiety was the support of a second carer who lived locally and was reliable. Carers from social classes IIIM (manual), IV and V were more likely to have this kind of support than carers from social classes I, II and IIIN (non-manual)., Conclusion: Although cancer patients from social class V were less likely than others to die in the hospice, social inequality in access to or utilisation of healthcare in terminal illness was not prominent in carers' accounts. When it did arise, it was associated with passivity in seeking information and support on the part of some carers from social classes IV and V. Carers from social classes IIIM-V received more regular and reliable support from their families than those from social classes I-IIIN.