A cohort of 3,916 Swedish copper smelter workers employed for at least 3 months between 1928 and 1967 was followed up through 1981. Arsenic exposure was estimated for different time periods at each workplace within the smelter. Detailed job records were linked to the exposure matrix, thus forming individual cumulative arsenic exposure measures for each smelter worker. Smoking history was collected for 107 lung cancer cases and 214 controls from the cohort. Lung cancer risks were positively related to cumulative arsenic exposure with smoking standardized relative risks ranging from 0.7 to 8.7 in different exposure groups. A negative confounding by smoking was suggested in the higher exposure categories. The interaction between arsenic and smoking for the risk of developing lung cancer was intermediate between additive and multiplicative and appeared less pronounced among heavy smokers.