Objectives: To examine the relative importance of broad social change, cohort-specific change, and population composition on trends in adult obesity over the past two decades. Methods: Using the National Health Interview Study from 1982 through 2002, 5-year birth cohorts are examined for differential trends in obesity. Logistic regression is used to separate out the effects of population composition from broad social change and cohort-specific change. Results: Results confirm that age-specific obesity rates have been increasing for successively born cohorts, indicating broad social change. There is little evidence for cohort-specific change, and only small effects of compositional change. Discussion: Although increasing diversity in the older population will probably result in higher rates of obesity in the future, increasingly sedentary lives and the uncertain impact of smoking cessation on weight outweigh population composition effects. More research is needed on the impact of lifestyle behaviors on the American population.