Aim To assess the efficacy in mortality prediction of a concordance of performance‐based (timed 10‐foot walk; performance‐oriented mobility assessment [POMA]) and self‐rated (reported ability to walk across a small room with no help from people or devices; activities of daily living [ADL]) assessments of mobility for Mexican Americans aged 75 years and older. Methods A longitudinal study of 2069 participants aged 75 years and older from the Hispanic Established Population for the Epidemiological Study of the Elderly wave 5 (June 2004 to January 2006) and wave 6 (February 2007 to February 2008) was carried out. Sociodemographic variables, performance‐based (timed 10‐foot walk) and self‐rated assessments (reported ability to walk across a small room without the help of any people or devices) of mobility, and mortality data were obtained. Results The ADL/POMA concordance assessment showed a prevalence of the “positively concordant” group (completed the walk and reported being able to walk, ADL and POMA both positive), followed by the “pessimist,” “optimist,” and “negatively concordant” groups at 80.09%, 10.50%, 3.78% and 5.63%, respectively. Logistic regression analyses showed that “negatively concordant” was a critical mortality predictor (OR 4.80; 95% CI 2.59–8.90) followed by “pessimist” (OR 1.94; 95% CI 1.12–3.36) as compared with the reference group, “positively concordant.” Conclusion The ADL/POMA concordance is an effective predictor of mortality for older Mexican Americans in the Hispanic Established Population for the Epidemiological Study of the Elderly. Geriatr Gerontol Int 2017; 17: 433–439.