SUMMARYIn the Kungsholmen project, a longitudinal study of a total population, samples of non-demented and demented elderly persons with a depressive disorder were followed over 3 years. Of 668 persons who received a full medical examination, 62 were found to have a depressive disorder. Of these, 28 were demented and 34 non-demented. Two-thirds of the 62 had not been previously recognized as having a depressive disorder. There was no difference in this regard between the demented and non-demented samples. After 3 years, only two out of 20 non-demented cases had recovered from their depressive disorder. This is in contrast to the demented cases, where only one of 11 were still depressed. These results indicate the need for greater attention to the recognition and management of depressive disorders in the elderly.KEY woms-Depression, dementia, elderly, longitudinal study.Knowledge of the outcome of depression in community samples is limited. Most of such information comes from clinic-based studies of treated series. The outcome variables have typically included the rate of recovery, the coexistence of depression with dementia and the progression of depression to dementia (Cole, 1990;Baldwin and Jolley, 1986;Baldwin, 1988;Gordon, 1981;Reding et al., 1985;Reifler et al., 1986;Henderson, 1990;Blazer et al., 1980;Post, 1972;Murphy, 1983;Burvill et al., 1989). However, clinic based studies have limitations. They cover only a subset of all the cases in the population. Cases that are unrecognized and untreated are excluded. Population-based studies are rare, but in a community study by O'Connor et al. (1990), 12/18 non-demented depressed cases were depressed 1 year later and two had developed a dementia. It is notable that in this study the sample excluded persons with good cognitive function and its follow-up period was relatively short.The present article uses data from the Kungsholmen study, which partly overcomes these limitations. This is a longitudinal community study that includes the full range of cognitive function with a follow-up time of 3 years. The aim of the present article is to examine the following issues:1. The recognition of depressive disorders in a community sample of elderly, with and without comorbid dementia 2. Whether the co-occurrence of dementia influences recovery 3. Whether depression progresses to dementia