This study evaluates the validity of subjective health measurement for racial/ethnic comparisons in the United States, by assessing whether allostatic load (AL) is equally associated with poor/fair self-rated health (SRH) for different racial/ethnic groups. This study used data from the National Health and Nutrition Survey (NHANES) for 2006-2010. Multivariable logistic regression models were fit and stratified by race/ethnicity to study the association between AL and poor/fair SRH. Higher levels of AL were associated with higher odds of reporting poor/fair SRH. However, this association differs by race/ethnicity. Analysis of interactions and racial/ethnic-stratified models suggest that AL is less associated with poor/fair SRH status for non-Hispanic Blacks and Hispanics populations. These results demonstrate that subjective health ratings potentially underestimate actual measures of biological health risk, especially for racial/ethnic minorities. As a result, population-based assessments of racial/ethnic health disparities based on SRH may be significantly understated.