This report describes the second in a series of three parallel cohort studies of asbestos factories in South Carolina, Pennsylvania, and Connecticut to assess the effects of mineral fibre type and industrial process on mortality from malignant mesothelioma, respiratory cancer, and asbestosis. In the present plant (in Pennsylvania) mainly chrysotile, with some amosite and a small amount of crocidolite, were used primarily in textile manufacture. Of a cohort of 4137 men comprising all those employed 1938-59 for at least a month, 97% were traced. By the end of 1974, 1400 (35%) had died, 74 from asbestosis and 70 from lung cancer. Mesothelioma was mentioned on the certificate in 14 deaths mostly coded to other causes. All these deaths occurred after 1959, and there were indications that additional cases of mesothelioma may have gone unrecognised, especially before that date. The exposure for each man was estimated in terms of duration and dust concentration in millions of dust particles per cubic foot (mpcf) from available measurements. Analyses were made both by life table and case referent methods. The standardised mortality ratio for respiratory cancer for the whole cohort was 105.0, but the risk rose linearly from 66.9 for men with less than 10 mpcf.y to 416.1 for those with 80 mpcf.y or more. Lines fitted to relative risks derived from SMRs in this and the textile plant studied in South Carolina were almost identical in slope. This was confirmed by case referent analysis. These findings support the conclusion from the South Carolina study that the risk of lung cancer in textile processing is very much greater than in chrysotile production and probably than in the friction products industry. The much greater risk of mesothelioma from exposure to processes in which even quite small quantities of amphiboles were used was also confirmed.