We investigated exposure-response relations for silicosis among 134 men over age 40 who had been identified in a previous community-based random sample study in a mining town. Thirty-two percent of the 100 dust-exposed subjects had radiologic profusions of small opacities of I/O or greater at a mean time since first silica exposure of 36.1 years. Of miners with cumulative silica exposures of 2 mg/m3-years or less, 20% had silicosis; of miners accumulating > 2 mg/m3 years, 63% had silicosis. Average silica exposure was also strongly associated with silicosis prevalence rates, with 13% silicotics among those with average exposure of 0.025-0.05 mg/m3, 34% among those with exposures of > 0.05-0.1 mg/m3, and 75% among those with average exposures > 0.1 mg/m3. Logistic regression models demonstrated that time since last silica exposure and either cumulative silica exposure or a combination of average silica exposure and duration of exposure predicted silicosis risk. Exposure-response relations were substantially higher using measured silica exposures than using estimated silica exposures based on measured dust exposures assuming a constant silica proportion of dust, consistent with less exposure misclassification. The risk of silicosis found in this study is higher than has been found in workforce studies having no follow-up of those leaving the mining industry and in studies without job title-specific silica measurements, but comparable to several recent studies of dust exposure-response relationships which suggest that a permissible exposure limit of 0.1 mg/m3 for silica does not protect against radiologic silicosis.