The cause-specific mortality was followed through 1981 in a cohort of 3,916 male Swedish smelter workers employed for at least 3 months from 1928 through 1967. Arsenic levels in the air of all workplaces within the smelter were estimated for three different time periods. Using this exposure matrix and detailed information of the work history, cumulative arsenic exposure could be computed for each worker. Standardized mortality ratios (SMRs) were calculated for several dose categories using age-specific mortality rates from the county where the smelter was situated. A positive dose-response relationship was found between cumulative arsenic exposure and lung cancer mortality with an overall SMR of 372 (304-450, 95% confidence interval). The lung cancer mortality was related to the estimated average intensity of exposure to arsenic but not to the duration. No positive dose-response relationship was found between arsenic and ischemic heart disease or cerebrovascular disease. There was also no evident dose-response relationship between estimated exposure to sulfur dioxide and lung cancer.