Female lung cancer mortality in France has risen by 3% annually over the past twenty years, reaching 4,500 deaths in 2000. This increase, which has not been observed among men (15,000 deaths in 2000), is attributed to the rise in smoking among women. To address the concerns of healthcare stakeholders, it seemed necessary to estimate, for this cancer, the mortality rates and the number of expected deaths, separately for men and women, over the next fifteen years in metropolitan France and in each of its regions. The analysis focused on the number of deaths from lung cancer from 1975 to 1999 and on estimated past and future population figures from 1975 to 2014, at the national and regional levels. Deaths and mortality rates for 1975–1999 were calculated by period and five-year age groups and projected for the period 2000–2014, for each region and for metropolitan France. The analysis uses the Bayesian approach of an age-period-cohort model with autoregressive constraints. In metropolitan France, male and female lung cancer mortality rates increased by 0.8% and 3.0% per year, respectively, between 1975 and 1999. During 1995–99, the age-standardized rate and the average annual number of deaths were, respectively, 78.9 per 100,000 and 20,600 for men, and 11.4 per 100,000 and 4,000 for women. For men, the largest increase was observed in Languedoc-Roussillon (+31%), and the smallest in Limousin (+3%). For women, the rate increased in all regions, but the largest increase was observed in Corsica (+314%), and the smallest in Auvergne (+37%). For metropolitan France, the age-standardized rate is projected to reach 78.1 and 75.7 per 100,000 for men in 2000–04 and 2010–14, respectively, representing a 3% decrease between these two periods. It is expected to reach 14.1 and 22.5 per 100,000 in 2000–04 and 2010–14, respectively, among women, representing a 60% increase between these two periods. At the regional level, for men, the results are mixed, ranging from a maximum increase in Aquitaine and Poitou-Charentes (17%) to a minimum in Burgundy (1%), and from a maximum decrease in Corsica (-15%) to a minimum in Lorraine (0%). For women, the largest increase is expected in Languedoc-Roussillon (107%), and the smallest in Nord-Pas-de-Calais (40%). The Bayesian approach to the age-cohort model is being used more and more frequently because it ensures the stability of rate projections and eliminates the need to analyze the etiological factors of cancer. It would, however, be interesting to supplement this model with a component that accounts for smoking, allowing for the development of scenarios based on declining consumption. (R.A.)