To improve the estimates of occupational fatality rates for persons employed in the construction industry, several sources of data on the number of fatalities (the numerator) and the number of persons engaged in construction work (the denominator) were examined. Based on this examination, the Census of Fatal Occupational Injuries (CFOI), compiled by the Bureau of Labor Statistics (BLS), was used to obtain numerators and the Current Population Survey, conducted by the Bureau of the Census for the BLS, was used to obtain denominators. Adjustments were made in the numerator to include only occupations that were included in the denominator. Occupations were divided into two groups-those in the construction trades and those in other occupations within construction (e.g., clerical, sales). The analysis found fatality rates of 14.2 and 13.3 per 100,000 person-years, respectively, for 1992 and 1993, with wide variation in rates among the different trades. There were also major differences among the trades in the types of fatal injuries. Self-employed workers had much lower death rates overall than wage workers, but this is largely due to much lower proportions of high hazard trades among the self-employed. There have been wide variations in the occupational fatality rates reported for construction workers each year due to the differing methods of estimating the number of fatalities by the different data sources. This study provides a baseline of fatality rates using the best available current data. It compares the results from these data sources with those from other sources that have been used and discusses some of the problems inherent in the data from other sources. This study provides a significantly improved protocol for the calculation of fatality rates against which later rates can be compared consistently. Nevertheless, many deficiencies in the data sources used are identified. There remains ample room for continued improvement.