To study the relation between exposure to crystalline silica and silicosis mortality. Although mortality is an important endpoint for regulators, there have been no exposure-response studies for silicosis mortality, because of the relative rareness of silicosis as an underlying cause of death, and the limited availability of quantitative exposure estimates. Data from six occupational cohorts were pooled with good retrospective exposure data in which 170 deaths from silicosis were reported. Standard life table analyses, nested case-control analyses, and risk assessment were performed. The rate of silicosis mortality in the combined data was 28/100 000 py, increasing in nearly monotonic fashion from 4.7/100 000 for exposure of 0-0.99 mg/m(3)-years to 233/100 000 for exposure of >28.1 mg/m(3)-years. The estimated risk of death up to age 65 from silicosis after 45 years of exposure at 0.1 mg/m(3) silica (the current standard in many countries) was 13 per 1000, while the estimated risk at an exposure of 0.05 mg/m(3) was 6 per 1000. Both of these risks are above the risk of 1 per 1000 typically deemed acceptable by the US OSHA. The findings from this pooled analysis add further support to the need to control silica exposure and to lower the occupational standards. Our estimates of lifetime silicosis mortality risk are probably underestimates as, in addition to exposure misclassification, our study might have suffered from outcome misclassification in that silicosis deaths might have been coded to other related causes, such as tuberculosis or chronic obstructive pulmonary disease.