The existing literature on educational inequalities in adult smoking has focused extensively on differences in current smoking and quitting, rather than on differences in never smoking regularly (initiation) by education in the adult population. Knowing the relative contribution of initiation versus quitting is critical for understanding the mechanisms that produce educational gradients in smoking because initiation and quitting occur at different points in the life course. Using data from 31 waves of the U.S. National Health Interview Survey (N = 587,174), the analyses show the relative likelihood of being a never versus former smoker by education, sex, and age from 1966 to 2010 and for birth cohorts from 1920 to 1979. The analyses also describe differences in the cumulative probability of quitting over the life course, and the role of initiation versus quitting in producing educational gaps in smoking. The results show that educational gaps in never smoking explain the bulk of the educational inequality in adult smoking. Differences in former smoking play a small and decreasing role in producing these gaps. This is true across the life course, whether measured at age 25 or age 50, and for both men and women. While the prevalence and age patterns of former smoking by education converge across birth cohorts, differences in never smoking by education increase dramatically. At the population level, educational gaps in adult smoking are produced by the combination of inequalities in initiation and quitting, with differences in initiation playing a larger role in producing the observed gaps. The portion of the gap explained by differences in quitting is itself a function of educational differences in initiation. Thus, educational gradients in adult smoking are tethered to experiences in adolescence. These findings have important implications for both understanding and addressing disparities in this important health behavior.