A logistic regression model was applied to 627 elderly men and women, who participated in all three data collections of the SENECA study in 1989, 1993, and 1999, to test the hypothesis that nutritional status in 80-85 year old persons was related to functional and cognitive status, but not significantly affected by living arrangement. Additionally, the authors hypothesized that relationships between cognitive status and self-care ability, between self-care ability and living situation, and between cognitive status and living situation would be stronger with increasing age. Nutritional status was categorized as being well nourished (> or = 24) or at nutritional risk (<24) using the 18-item mini-nutritional assessment questionnaire (MNA). Diminished self-care ability was defined as inability to perform all 7 self-care items on the ADL instrument. Respondents were deemed to have possible cognitive impairment if they scored <24 on the MMSE test. Residence situation was categorized as either living alone, with spouse/partner, or with others. Nutritional risk was found to be associated with diminished cognitive status and diminished self-care ability, but not associated with living alone. Elderly people with diminished cognitive function and diminished self-care ability had a more than two times higher risk of being at nutritional risk. In addition, the strength of relationships between cognitive status and self-care ability, self-care ability and living situation, and cognitive ability and living situation all increased over time. In the oldest old lifestyle characteristics and functional ability appeared to be stronger predictors of risk for malnutrition than in younger adults.