It has been hypothesized that beta-carotene mediates the association between low serum cholesterol levels and increased risk of lung cancer. It follows from this assumption that this association should be greater in population strata with a low intake of beta-carotene than in with those with a high intake. To investigate this hypothesis, we analysed dietary beta-carotene, plasma beta-carotene and serum cholesterol levels in 20 male smokers with lung cancer and 103 male smoking controls, a subsample taken from a larger case-control study on oxidant-antioxidant status. As predicted, we found that the association between low serum cholesterol levels and lung cancer risk was greater in subjects with low plasma beta-carotene. Controlling for plasma beta-carotene decreased but did not negate the magnitude of the inverse association between serum cholesterol and lung cancer. A low serum cholesterol level tended to increase the risk associated with low plasma beta-carotene. Our data suggest that a low plasma beta-carotene does not totally explain the association between serum cholesterol and lung cancer.