Epidemiologic studies of workers exposed to silica were reviewed to identify data on airborne concentrations of quartz that are not associated with an increased risk of silicosis, the lowest concentrations associated with silicosis, and studies that used statistical models to quantitate the risk of silicosis as a function of silica exposure. The no observed adverse effect levels varied from 7 to 100 mu g center dot m-3, and the lowest observed adverse effect levels ranged from 8 to 252 mu g center dot m-3 in five different cohorts. Studies using quantitative exposure-response models revealed a wide difference in the cumulative risk estimates for silicosis. The differences in the risk estimates and the no observed and lowest observed effect levels may have been the result of errors in exposure estimates, physicochemical characteristics of silica and quartz content of the dust, cohort differences, and reader variability. Further research is needed to define the dose-response relationship between silica exposure and silicosis.