In this study, a number of demographic, health, and personality measures were collected from a sample of institutionalized and non-institutionalized elderly people in Newfoundland. These measures were then regressed on the number of medicines used. The analyses showed that the primary predictor of medicine use in both samples is Disease Severity, a combination of the number of different diseases or disorders and the extent of their severity. This factor accounts for about 28 to 33 per cent of the variance. No other factor predicted medicine use for institutionalized elderly, but for non-institutionalized elderly, Health Rating, how the individual perceives his or her own health, accounts for an additional 5 to 7 per cent of the variance. Disease Severity and Health Rating do not interact, but make independent, additive contributions to predicting medicine use. Disease Severity and Health Rating also predict total drug use and changes in drug use over a period of 12 to 18 months. When variance due to Disease Severity and Health Rating are accounted for, sex and age do not significantly predict medicine use.