OBJECTIVE: To assess whether the association between cognitive ability (IQ) and early mortality is mediated by socioeconomic status (SES) or whether the association between SES and mortality reflects a spurious association caused by IQ. METHODS: The participants were from the US National Longitudinal Survey of Youth (n = 11,321). IQ was assessed at age 16 to 23 years and the participants were followed up to 40 to 47 years of age. RESULTS: Controlling for sex, birth year, race/ethnicity, baseline health, and parental education, higher IQ was associated with lower probability of death (odds ratio (OR) per 1-standard deviation increase in IQ = 0.78, 95% confidence interval (CI) = 0.66, 0.91). This association disappeared (OR = 0.99, 95% CI = 0.81, 1.20) when adjusted for education and household income. Adjustment for IQ had no effect on the association between SES and mortality. These findings were similar in Hispanic, Black, and White/other participants and in women and men. Parental education moderated the IQ-mortality association so that this association was not observed in participants with low parental education. CONCLUSIONS: Low IQ predicts early mortality in the US population and this association is largely explained by SES. The results do not support the alternative hypothesis that the socioeconomic gradient in early mortality would reflect IQ differences.