Purpose: This study investigates gender differences in the association between self-rated health (SRH) and mortality. This association has been well-documented, but findings regarding gender differences are inconsistent. The specific objectives were (a) to examine these differences in a short and a long time frame, (b) to examine these differences among old and old-old people, and (c) to address the question of whether this association is based on the accuracy of poor SRH as a predictor of future decline, and/or of better SRH as a predictor of longevity. Design and Methods: The study is based on an Israeli nationally representative sample of 622 women and 730 men who were interviewed about their SRH, as well as sociodemographic information and other measures of health, physical functioning, cognitive status, and depression. Results: For both genders, SRH was associated only with shorter term mortality (within the next 4 years) and not with longer-term mortality (9 years of follow-up). This association was strongest among the old (ages 75-84) women, compared with the old men and with the old-old (85-94) women and men. A possible explanation may be related to differences in the accuracy of excellent SRH at very old age. Implications: The SRH-mortality association may differ among age and gender groups. Identifying the conditions under which it is more accurate will enable researchers and practitioners to know when it can be utilized. It is important to assess differences in the accuracy of poor SRH as well as of excellent SRH as predictors of future health outcomes.