Two cohorts of Quebec chrysotile miners and millers were selected in order to study how radiological changes in asbestos workers predict mortality. Cohort "A" comprised 988 men, with dust exposure, smoking history, respiratory symptoms and lung function recorded in 1967- 68, and six independent readings of an earlier chest radiograph; 130 men had died before 1976. Cohort "B" comprised all 4,559 men not in Cohort "A" who were born 1891 through 1920, with at least a month's employment and an x-ray; dust exposure, smoking history and a single x-ray reading were obtained. By the end of 1975, there had been 1,455 deaths in Cohort "B". A high Relative Risk (RR) of total mortality was related to each radiographic feature, cigarette smoking, excessive dust exposure, and, in Cohort "A", dyspnoea and poorer function. Death from pneumoaonioois was associated with small irregular parenchymal opacities, of profusion usually greater than 1/0, and with heavy dust exposure, but not with smoking. Almost all who died from chest cancer had smoked cigarettes (very high RR), have been heavily exposed to asbestos (high RR), or both. About half their x-rays showed pneumoconiotic change. There was insufficient evidence to determine whether chest cancer was a complication of asbestosis or an independent reaction to asbestos exposure. Five of the six mesothelioma cases had positive but varied radiographic appearances. The contribution of asbestos dust to some deaths attributed to heart disease was substantiated by high RRs for severe exposure, and for small parenchymal opac ities; smokers, and the breathless, also had high risks. Other neoplasms showed no consistent dust or x-ray patterns. RRs of deaths from some other causes were high for certain radiographic features. Not all these results can yet be explained. The main findings validated the UICC/Cincinnati classification convincingly. Studies to permit generalization are recommended.