The nature of the interaction between smoking and occupational exposure is controversial, in part because of lack of agreement on the definition of interaction and in part because of the scarcity of adequate epidemiologic data. Occupational investigators have assessed interaction primarily as a departure from an additive rather than from a multiplicative model of relative risks (or rate ratios). To determine whether smoking modifies the effect of occupational lung carcinogens, the literature was reviewed for the only four established occupational lung carcinogens for which there are data on smoking: radon daughters, asbestos, arsenic, and chloromethyl ethers. Where possible, departure was assessed from both an additive (synergism) and a multiplicative model (effect modification). Only nine studies were considered to have sufficient sample size and to provide sufficient information on tobacco use and occupational exposure to evaluate interaction. The existing data were contradictory for three of the agents studied: asbestos, radon daughters, and arsenic. Inconclusive or contradictory findings may result from small sample size or lack of comparability of the level of occupational or tobacco exposure. It is noteworthy that, for these four agents, whenever smoking did modify the effect of occupational exposure, the lung cancer rate ratio was greater for nonsmokers (compared to nonexposed nonsmokers) than smokers (compared to nonexposed smokers). However, with the exception of chloromethyl ethers, absolute lung cancer rates were higher for smokers than nonsmokers, regardless of occupational exposure.