The 1983 study of dependency of subjects in institutional care in Dunedin was repeated two years later. A significant increase in levels of dependency in residential homes, particularly in the Religious and Welfare sector was found. In 1983 there were 29 high dependency residents and 73 medium dependency residents in residential homes. In 1985 these numbers had increased to 55 and 86 respectively. There was no change in the number of low dependency residents. In 1983, 6 high dependency residents had been admitted to residential home care in the year prior to the study. In 1985 the number of high dependency residents recently admitted had increased to 23. There had also been a significant increase in the dependency of patients in Religious and Welfare continuing care hospitals. Of the 933 subjects in institutional care in 1983 who were able to be followed, 354 (37.9%) died in the following 2 years. Mortality rate was higher for those in hospital care (48.1%) than for those in residential home care (29.6%). Mortality rates were higher in more dependent subjects and this was evident for each measure of dependency.