The mortality of a cohort of 1487 male patients with silicosis in a population-based register followed up from 1980 to 1986 was evaluated with reference to the mortality rates of the general male population. A striking excess of deaths from all causes (observed 368, standardized mortality ratio, SMR 3.00) was noted. Seventy-four percent of the deaths were due to respiratory conditions and complications directly or indirectly related to silicosis. The risk of death was especially higher than expected in younger patients under 45 years of age. Patients with simple silicosis of profusion category 1 did not appear to be at any increased risk of death relative to the general population, but increasing excesses of death were associated with greater extent of simple and conglomerate disease. These increased mortality risks were observed in tuberculosis-free patients as well as in those who never smoked. For the same extent of silicotic disease, the risk of death was higher if tuberculosis occurred. There was no evidence, however, that patients who smoked were more likely to suffer a worse mortality outcome than those who did not. Age at diagnosis, extent of silicotic disease and the occurrence of tuberculosis were therefore strong predictors of mortality in patients with silicosis.