The purpose of this paper was to identify predictors of social isolation and loneliness for very old rural and urban adults. With data from the 1996 Aging in Manitoba Study (N = 1,868; age range 72-104), separate multiple regression models were constructed for rural and urban sub-samples, using the life space index (LSI) to measure social isolation as one outcome, and a loneliness index created by the authors from a combination of items to measure loneliness as a second outcome. Different factors were found to predict the outcomes for the two sub-samples. The models with isolation as the outcome produced five predictors for the rural sub-sample and three for the urban sub-sample. Only living alone was the same for both groups. The models with loneliness as the outcome produced five predictors for the rural sub-sample and two for the urban sub-sample, again with only one factor in common--four or more chronic illnesses. We conclude that health and social factors are important predictors of social isolation and loneliness, and sensitivity to these factors may improve the experience of older adults.