We sought to describe the changing death rates from lung cancer in the US white population in sequential birth cohorts, adjusting for cohort smoking prevalence and duration. We searched the US mortality database (1960-1994) for all deaths among whites in which lung cancer was listed as the underlying cause of death. To determine the population at risk for lung cancer, we used the 1970, 1978-1980, and 1992 National Health Interview Surveys to estimate the annual number of current and recent smokers (those who had quit within 5 years) in 11 5-year birth cohorts, starting in 1901. We then determined annual lung cancer mortality rates for each birth cohort, stratified by sex and adjusting for the prevalence and duration of smoking. The population-based rates of lung cancer mortality were much higher among men than among women across all ages and birth cohorts, reflecting higher smoking rates among men. These differences decreased after we controlled for current and recent smoking within the cohorts and were slightly increased in women after we controlled for duration of smoking. Differences in lung cancer death rates across birth cohorts of US men and women primarily reflect differences in the prevalence and duration of smoking in these birth cohorts. Changes in cigarette design that have greatly reduced tar yields have a relatively small effect compared with that of people's smoking status and duration of smoking.