Our objective is to describe lung cancer mortality trends in Andalusia, Spain, during the period 1975-2004 using age-period-cohort analysis (APC). For each gender, age-standardised (European Standard Population) mortality rates from lung cancer were computed based on the causes of death on the death certificates from the official registry of vital statistics in Andalusia. In men, after climbing considerably from 1975 to 1994, adjusted lung cancer mortality rates, have been declining by 0.8% per year since 1994. For women, the mortality from lung cancer was almost constant but tended to increase after 1994 (average annual increase of 2.1%). Among males, the cohort effect was steadily and appreciably upwards to the cohort born around 1905, then levelled off, and declined in the youngest generations. An increasing period effect was also observed until 1995. For females, cohort values decreased until the cohort born around 1930, then levelled off, and increased for women born since 1940. Period effect trend was downward until 1990, and upward thereafter. In conclusion, the cohort effect observed for women born since 1940 suggests the start of a lung cancer epidemic associated with a higher prevalence of smokers in women. The decrease in prevalence of smokers among males and the decrease in mortality in younger age groups suggest that the trend initiated in 1994 will continue as long as smoking prevalence continues to decrease.