The usual method for estimating population exposure to cigarette tobacco has been annual per capita consumption of cigarettes, expressed as pack-years or numbers of cigarettes consumed. This technique is shown to result in an inaccurate estimate of exposure. It underestimates by 11 years the latency period from peak exposure to peak mortality. Over the years, cigarettes changed markedly. Filters came into prominence and tobacco was "fluffed". On average, tobacco content of cigarettes decreased 39.1% from 1953 to 1981. National per capita consumption of cigarette tobacco declined by 43%. Total exposure to cigarette tobacco has been declining for males for approximately 35 years; for females for 20 years. As of 1982, the secular trend for lung cancer mortality was declining for women below age 45 and for most age groups of men below 65. We appear to be at the threshold of a reversal in overall lung cancer mortality.