Objective: Self-rated health (SRH) has been shown to be related to mortality in many countries, even after controlling for variations in objective health status. A notable problem with subjective measures such as SRH is the presence of different reporting styles, as demonstrated by a great level of variation in ratings of the same hypothetical cases (anchoring vignettes). It is unclear how such reporting heterogeneity may affect the predictive utility of SRH. This study intends to estimate the predictive utility of SRH before and after reporting style differences are accounted for. Methods: Data were from the 2012 and 2014 waves of the nationally representative China Family Panel Studies. Respondents were asked to rate their own health as well as the health status of two hypothetical persons (i.e., anchoring vignettes) using a 1-5 Likert scale (poor, fair, good, very good, excellent) in CFPS2012. We formed an adjusted SRH score based on the two anchoring vignettes and assessed whether it predicted respondents’ mortality status in CFPS2014. The predictive utility of the adjusted score was compared with that of the raw SRH. Analysis were restricted to those aged 65 and above. Results: The raw measure of SRH is related to mortality (OR=0.76, 95% CI=0.63, 0.91), even after accounting for age, gender, marital status, education, family income, chronic disease and in-patient visit in the past year. The predictive utility did not improve (OR=0.82, 95% CI=0.69, 0.98) when using the adjusted ratings of SRH accounting for possible differences in reporting styles. Conclusion: SRH is predictive of mortality in the next two years among Chinese older adults aged 65+ even after controlling for sociodemographic variables and a few objective indicators of physical health. Adjusting for reporting styles in SRH does not increase the predictive utility in predicting old age mortality in the next two years.