Three simple models for the asbestos-smoking interaction on human lung cancer production are considered. In the first model the excess incidence of lung cancer independently due to asbestos and to smoking adds together when both agents are present (additive model). In the second the addition of each one of the two agents produces an effect (increase in lung cancer incidence) which is proportional to the effect of the other (multiplicative model). In the third, asbestos can only increase lung cancer incidence in the presence of smoking. As previously found by other investigators, the additive model appears the least plausible in the light of the data from two published epidemiological studies. A discrimination between the other two models is attempted through a detailed analysis of the five published epidemiological studies today available which provide information on occupational asbestos exposure, smoking habits and lung cancer risk. Although the data do not allow a definitive discrimination, the multiplicative model appears to be more plausible, being also consistent with a multi-stage carcinogenic mechanism and with evidence from animal (rat) experiments. It is relevant both for biology and for public health that in this model asbestos and smoking are regarded as independently capable of producing lung cancer in humans and that they act synergistically when exposure to both occurs.