A number of epidemiological studies have reported associations of beta-carotene plasma levels or intake with decreased lung cancer risk. However, intervention studies in smokers have unexpectedly reported increased lung tumor rates after high, long-term, beta-carotene supplementation. Recently, detailed analyses by stratification for smoking habits of several large, long-term intervention or epidemiological trials are now available. The ATBC study, the CARET study, the Antioxidant Polyp Prevention trial, and the E3N study provide evidence that the adverse effects of beta-carotene supplementation are correlated with the smoking status of the study participants. In contrast, the Physician Health Study, the Linxian trial, and a pooled analysis of 7 epidemiological cohort studies have not supported this evidence. The ferret and A/J mouse lung cancer model have been used to investigate the mechanism of interaction of beta-carotene with carcinogens in the lung. Both models have specific advantages and disadvantages. There are a number of hypotheses concerning the beta-carotene/tobacco smoke interaction including alterations of retinoid metabolism and signaling pathways and interaction with CYP enzymes and pro-oxidation/DNA oxidation. The animal models consistently demonstrate negative effects only in the ferret, and following dosing with beta-carotene in corn oil at pharmacological dosages. No effects or even protective effects against smoke or carcinogen exposure were observed when beta-carotene was applied at physiological dosages or in combination with vitamins C and E, either as a mixture or in a stable formulation. In conclusion, human and animal studies have shown that specific circumstances, among them heavy smoking, seem to influence the effect of high beta-carotene intakes. In normal, healthy, nonsmoking populations, there is evidence of beneficial effects.