Three parallel cohort studies of asbestos factory workers were undertaken to investigate the effects of mineral fibre type and industrial process on malignant mesothelioma, respiratory cancer, and asbestosis. This report describes the mortality of a cohort of 2543 men, defined as all those employed for at least a month from 1938 to 1958 in a textile plant in South Carolina in which chrysotile was the only type of asbestos used. Of these, 863 men (34%) had died before 31 December 1977, one from malignant mesothelioma. Twenty one deaths were ascribed to asbestosis and 66 to cancer of the lung. Compared with the number expected from South Carolina, there was an excess of 30 deaths from respiratory cancer (ICD 160-164) in men 20 or more years after first employment (SMR 199.5). In men employed five years or more, no SMRs for this category rose above 300. Individual exposures were estimated (in mpcf X years) from recorded environmental measurements. Life table analyses and "log-rank" (case-control) analyses both showed a steep linear exposure-response that was some 50-fold greater at similar accumulated dust exposures than in Canadian chrysotile mining and milling. These findings agree closely with those from another study in this plant and confirm that mesothelioma is rarely associated with chrysotile exposure. Cigarette smoking habits did not greatly differ between the textile workers and the Canadian miners and millers. The far greater risk of lung cancer in the textile industry, if not attributable to other identified cocarcinogens, may be related to major differences in the size distribution of fibres in the submicroscopic range which are not detected by the usual fibre or particle counting procedures.