The general health questionnaire (GHQ) and its relation to hospital care and risk of death were studied prospectively in the total elderly population living in a community in northern Finland. Of the eligible elderly, 982 or 85% took part in the study, returning the GHQ. A high score on the questionnaire was considered an indicator of minor psychiatric illness. It was found that the average number of hospital days, calculated per person-years, was twice as high in the high scorers than in the low scorers. When the effects of age and certain diseases were controlled for, the high scoring men and women had a significantly higher risk of being hospitalized for at least ten days over a five-year period than the 0-1 scorers. The risk ratio was 1.6 and 1.4 for men and women, respectively. The ten-year mortality was significantly higher in the high scoring men and women than in the low scorers, when the effects of age and certain diseases were controlled for. The risk ratios for both men and women were 1.4. The GHQ was found to predict hospital care, physician visits, home services and mortality in a representative group of rural elderly people.