Incidence rates of lung cancer have been markedly lower for Fiji than for other South Pacific countries, despite similar rates of smoking. We conducted population-based surveys in several island nations of the South Pacific (Cook Islands, Fiji, Tahiti and New Caledonia) and used data from Caucasian, Japanese, Hawaiian, Filipino and Chinese controls in a case-control study of lung cancer in Hawaii to investigate the role of diet in explaining differences in lung cancer incidence among 20 ethnic-sex groups. In a stepwise linear regression of lung cancer rates on smoking, diet and other variables, smoking, as expected, explained the majority (61%) of the variability in incidence. However, several dietary components also explained significant portions of the variance. Lutein intake explained 14% and vitamin E intake, cholesterol intake and height explained 5-7% each of the remaining variance in incidence. Associations with lutein and vitamin E were inverse, whereas those with cholesterol and height were direct. Dietary beta-carotene intake was not associated with lung cancer incidence. These ecological data provide evidence for a protective effect of lutein against lung cancer. A protective effect of dietary vitamin E and a risk-enhancing effect of dietary cholesterol are also suggested.