Survival analyses are used to examine the predictive value of spouse-rated limitations due to health problems and spouse-rated life expectancy for 3-year mortality in a communication sample of elderly Australian married couples, net of health status indicators, health behaviors, and sociodemographic factors. Tests for gender differences in the effects and for empirical overlap with self-ratings of health expectancy were also done. Findings show that both spouse-ratings are predictive of husbands' mortality, while for wives' mortality the parallel measures are not predictive in the adjusted model. Wives' rating are not substitutes for other health indicators or for self-ratings of health and life expectancy. The independent effect suggests that wives are more astute reporters or judges of their husbands' mortality risk for that wives' perceptions indicate tangible influences on husbands' health and mortality risk. Future analyses should be directed toward finding mechanisms linking spouse perceptions to health outcomes.