The usefulness of Mine Safety and Health Administration (MSHA) compliance data for epidemiologic research was examined to determine if the dust exposure data could be used to establish quantitative exposure response relationships for the NIOSH National Study of Coal Workers' Pneumoconiosis (NSCWP). The MSHA database consisted of approximately 18 years worth of personal exposure monitoring of United States underground coal mines and contained approximately 4.7 million samples. The strategy for examining the data consisted of choosing an appropriate sample of mines, performing a descriptive analysis of the distribution of data points and mean exposure values within each mine, year, and job category, assessing bias, reducing the number of categories for analysis, specifying the best model for predicting mean exposures, and calculating estimated exposures. A preliminary analysis of samples from two mines in the original NSCWP study showed that the data were highly skewed. The distribution after a logarithmic transformation was more normal, having a skewness coefficient of -0.43. The data seemed to have an overrepresentation of samples having concentrations of 0.1mg/m3. The data showed a downward trend in dust exposures for the period 1971 through 1986; however, the year to year variability was large. The authors conclude that the data collected by the MSHA coal mine dust compliance program should be very useful for establishing relatively precise exposure estimates for the fourth and subsequent rounds of the NSCWP.