To explore the role of arsenic as a human carcinogen, the respiratory cancer mortality experience (1938 to 1977) of 8,045 while male smelter employees in Montana was examined relative to cumulative exposure to arsenic trioxide and was compared with that of the white male population of the same region. Exposure to arsenic was estimated for various work areas from industrial hygiene reports of average concentrations present in the smelter. Respiratory cancer mortality was analyzed further by time period of first employment and maximum lifetime exposure to arsenic trioxide. When exposure was estimated with arithmetic means of measured concentrations among men first employed prior to 1925, respiratory cancer mortality increased linearly with increasing cumulative exposure group, ranging from two to nine times expected; among those first employed in the period 1925 to 1947 it also increased linearly with increasing cumulative exposure group.