We use the U.S. National Longitudinal Survey of Youth 1979 and other sources to examine how cognitive performance near the end of secondary schooling relates to labour market outcomes through age fifty. Our preferred estimates control for individual and family backgrounds, non-cognitive attributes, and survey years. We find that returns to cognitive skills rise with age. Although estimated gains in lifetime incomes are close to those reported earlier, our preferred estimates make multiple offsetting improvements. Returns to cognitive skill are greater for blacks and Hispanics than for non-Hispanic whites, both in relative and absolute terms, with gains in work hours being more important than in hourly wages.