When risk factors are examined concurrently, they tend to show mixed results in predicting mortality among the old-old. The purpose of this study was to compare a set of physical predictors with a set of mental predictors, all considered as most common and predictive in the literature, to assess their relative dominance in predicting mortality at old-old age. Based on Baltes's incomplete architecture model and the disablement process, the authors postulated that physical predictors of mortality would diminish the impact of mental predictors of mortality. The database used for this study was the multidimensional survey of the Cross-Sectional and Longitudinal Aging Study conducted from 1989 to 1992 with a follow-up of mortality after 10 years. Participants (N = 1,369) were drawn from a national sample of the Jewish Israeli population aged 75 to 94. The mean age of the sample was 83.52 (SD = 5.42). Results of hierarchical Cox regression models showed that besides sociodemographic effects (mainly age, gender, and marital status), physical disability, physician visits, and medication consumption predicted mortality. Mental predictors (cognitive impairment and life evaluation) lost their predictive power when the physical predictors were introduced. Hence, biological and physical risk factors predominated over mental risk factors when examining prediction of mortality among the old-old age population.