Abstract Background The relationship between subjective and objective health is complex and not always matched. Although frailty and self-rated health (SRH) have been separately associated with adverse outcomes, their joint effects remained unclear. Methods Participants were 5 300 adults ≥60 years from the China Health and Retirement Longitudinal Study in 2011. Frailty, measured by the validated physical frailty phenotype approach, was classified as nonfrail, prefrail, and frail. SRH was categorized into 3 groups: excellent/very good/good, fair, and poor/very poor. We used the Cox models to examine the independent and joint association of frailty and SRH with mortality. We used the interaction approach to determine whether the association of SRH with mortality differed by frailty. Subgroup analyses were conducted by depression and cognitive impairment. Results About 8.1% of frail participants reported excellent/very good/good health; 21.2% of the nonfrail reported poor/very poor health. Prefrailty and frailty were associated with a 1.63- and 2.38-fold increase in the hazard of mortality than the nonfrail, respectively, after adjusting for SRH. Reporting fair and poor/very poor health was associated with a 29% and 100% increase in the hazard of mortality, respectively, after adjusting for frailty. No significant interaction was found. Prefrail and frail older adults with excellent/very good/good health had a similar mortality as the nonfrail with poor/very poor SRH. The association of SRH with mortality was less pronounced among individuals with depression or cognitive impairment. Conclusions SRH is a potential marker of resilience among people living with frailty that may be a target for ameliorating health risks induced by frailty.