Trends in death certification rates from lung cancer in broad Italian geographical areas (north/centre/south) were analysed over the period 1969-1987. In northern Italy, lung cancer rates in young and middle-aged males reached a peak between the mid and late 1970s, and tended to decline afterwards; only above age 60 was mortality still rising in the 1980s. A similar pattern of age-specific rates was observed in central areas, while in the South rates tended to level off in the early 1980s only below age 55, but were still upwards in subsequent age groups. Consequently, the north/south ratio for the overall age-standard rate increased slightly between the late 1960s and mid 1970s, from 1.68 (corresponding to a world standardised rate of 47.1/100,000 in the north vs. 28.1 in the south) to 1.73, but declined to 1.55 between 1985 and 1987 (for a rate of 69.1/100,000 males in the north vs. 44.6 in south). In the younger age groups a diverging pattern was observed: at ages of 25-34 rates in 1985 and 1987 were apparently higher in the south (1.0 vs. 0.9/100,000 in the north), and in the 35-44 age group the north/south ratio decreased from 1.7 to 1.2 (with rates of 12.9 and 10.7, respectively, in 1985 and 1987). Among females, lung cancer rates increased in all geographical areas and age groups except the youngest (25-34 years). Under the age of 50, the rises were proportionally similar in various geographical areas, thus widening the north/south difference in absolute terms. Above the age of 50, the north/south difference tended to be wider in relative terms too, reaching a factor of 2 in the 65-74 age group. The overall age-standardised north/south ratio for females increased from 1.51 in 1969-1974 (5.6 vs. 3.7/100,000) to 1.87 in 1985-1987 (8.4 vs. 4.5/100,000). These trends reflect changes in smoking habits in subsequent generations of Italian males and females from different areas of the country, and confirm the central role of cigarette smoking in lung cancer rates in various populations, although this does not exclude some influence by other, mainly occupational, lung carcinogens on the substantial differences in lung cancer rates in various Italian geographical areas.