This paper provides an overview of the relationship between tobacco use and early cancer mortality. It presents a retrospective examination of trends in smoking behavior and how these trends affected the national lung cancer mortality pattern during this century. Information on smoking prevalence is presented for black and white men and women for each 5-year birth cohort between 1885 and 1969. The author argues that the lung cancer mortality pattern observed in the United States since 1950 is entirely compatible with changes in smoking behavior among the various birth cohorts examined. The paper also reviews our current scientific knowledge about the etiological relationship between cigarette smoking and site-specific cancer mortality, with particular emphasis on lung cancer. Data on other forms of tobacco use and cancer mortality risks are included as are data on environmental tobacco smoke exposures and nonsmokers' lung cancer risk. Data are presented to demonstrate that cigarette use alone will be responsible for nearly one-third of the U.S. cancer deaths expected in the United States in 1995, or 168,000 premature cancer deaths. Among males, 38% of all cancer deaths are cigarette related, while among women 23% of all cancer deaths are due to cigarettes. These totals, however, include neither the cancer deaths that could reasonably be attributed to pipe, cigar, and smokeless tobacco use among males nor the estimated 3000 to 6000 environmental tobacco smoke-related lung cancer deaths that occur annually in nonsmokers. It is concluded that tobacco use, particularly the practice of cigarette smoking, is the single greatest cause of excess cancer mortality in U.S. populations.