To compare the quality of care for spouses of dying people in St Christopher's Hospice, London and nearby hospitals in 1994 and to make comparisons with earlier studies of the same setting, interviews were undertaken with spouses of people who had died from cancer in these settings, matched by the deceased's age and sex. The subjects comprised 66 people whose spouses had died in 1994, 33 of whom had died in the hospice, 33 in local hospitals. The mean age was 69.3; 61% were female. No significant differences between groups were found on the outcome measures (adjustment to bereavement, anxiety and psychosomatic symptoms at the time of the final admission), largely replicating earlier studies in the same setting. In various respects hospitals have moved closer to hospice practice: in contrast to earlier studies, there were no differences in visiting patterns or in spouses helping with inpatient care. Bereavement follow-up is now initiated in some cases by hospitals. Liberal visiting hours could be stressful, however, and spouses sometimes helped with care to remedy staff shortcomings. Regret at not being present at the death was more common in the hospital group. Bereavement interventions may have more effect on subsequent adjustment if targeted on high-risk individuals. Other benefits than adjustment, however, may be achievable. Steps to create a calm atmosphere on hospital wards where people can be with their relatives near the time of death are desirable.