Beta-carotene (BC) was found to enhance lung cancer risk in smokers. This adverse effect was unexpected because BC was thought to act as an anti-oxidant against cigarette smoke-derived radicals. These radicals can directly or indirectly damage DNA, leading to the formation of pro-mutagenic DNA lesions such as 8-oxo-7,8-dihydro-2'-deoxyguanosine (8-oxo-dG) and 3-(2-deoxy-beta-D-erythro-pentafuranosyl)pyrimido[1,2-alpha]purin-10(3H)-one deoxyguanosine (M(1)dG). Later, it was suggested that high concentrations of BC could also result in pro-oxidant effects. Therefore, we investigated whether high but physiologically feasible concentrations of BC were able to alter (i) the formation of radicals in vitro assessed by electron spin resonance spectroscopy, (ii) the levels of 8-oxo-dG and M(1)dG in vitro in lung epithelial cells after incubation with hydrogen peroxide (H(2)O(2)) and the smoke-derived carcinogen benzo[a]pyrene (B[a]P) and (iii) the levels of 8-oxo-dG and M(1)dG in vivo in ferrets' lung after chronic exposure to B[a]P. BC increased in vitro hydroxyl radical formation in the Fenton reaction but inhibited the formation of carbon-centered radicals. Similarly, BC was able to enhance 8-oxo-dG in vitro in lung epithelial cells. On the other hand, BC significantly inhibited M(1)dG formation in lung epithelial cells, especially after induction of M(1)dG by H(2)O(2) or B[a]P. Finally, BC supplementation of ferrets also resulted in a significant decrease in M(1)dG, but in contrast to the in vitro experiments, no effect was observed on 8-oxo-dG levels, probably because of increased base excision repair capacities as assessed by a modified comet assay. These data indicate that the fate of BC being a pro- or anti-oxidant strongly depends on the type of radical involved.