Lung cancer mortality was examined in a cohort of 4393 men employed at a zinc-lead-cadmium smelter. There was an excess of lung cancer (overall SMR = 124.5, 95% confidence interval 107-144) which was particularly evident for those employed for more than 20 years. A statistically significant trend in SMRs with increasing duration of employment was apparent. Quantitative estimates of exposure to cadmium and ordinal rankings for lead, arsenic, zinc, sulphur dioxide, and dust were used to calculate cumulative exposures from job histories. Matched logistic regression was used to compare the cumulative exposures of cases of lung cancer to those of controls matched for date of birth and date of starting work and surviving at the time of death of the matched cases. The increasing risk of lung cancer associated with increasing duration of employment could not be accounted for by cadmium and did not appear to be restricted to any particular process or department. Although lung cancer mortality was associated with estimates of cumulative exposure to arsenic and to lead, it was not possible to determine whether the increased risk might be due to arsenic, lead, or to other contaminants in the smelter. These results are compared with findings from other non-ferrous smelters.