The interaction between occupational arsenic exposure and tobacco smoking and its relationship to lung cancer mortality among 228 deceased Swedish copper smelter workers was studied with the case-referent technique. Arsenic exposure was assessed via detailed company records, and information on smoking habits was gathered from the next of kin. The age standardized rate ratio for death from lung cancer was 3.0 for arsenic-exposed nonsmokers and 4.9 for smokers without occupational arsenic exposure in relation to nonarsenic-exposed nonsmokers. For arsenic-exposed smokers the rate ratio was 14.6, indicating a multiplicative effect of the two exposures. Eighty-five percent of all deaths from long cancer among the smelter workers could be "explained" by arsenic exposure and/or smoking. The interaction between arsenic and smoking suggests that a strong preventive effect on lung cancer incidence could be obtained by decreasing either one of the exposures or by disaggregating them.