This study examined the hypothesis that sociodemographic characteristics such as age, education, race, and gender would be predictive of Multidimensional Health Locus of Control Subscale scores in a population-based sample of 342 community dwelling elderly individuals. Bivariate analysis revealed associations between black race, lower socioeconomic status, and lower education on the Chance and Powerful Others Subscales. While the multivariate analysis revealed no predictors for the Internal Subscale, a higher socioeconomic status, white race, and a higher level of education continued to predict low scores on the Chance Subscale when controlling for all other variables. Scores on the Powerful Others Subscale appeared to be a function of socioeconomic status and gender. Of note, the higher the education level for both men and women, the lower the scores on the Chance and Powerful Others Subscales. This sex by education interaction term reached statistical significance for the Chance Subscale. The results demonstrate the measurable influence of sociodemographic variables on the health beliefs of community dwelling elderly individuals.