Retrospective dose assessment is required for radioepidemiologic studies conducted at the National Insitute for Occupational Safety and Health (NIOSH). Specific examples of dose reconstruction methods are presented from recent epidemiologic studies of nuclear workers employed at the Idaho National Engineering and Environmental Laboratory (INEEL), and Portsmouth Naval Shipyard (PNS). Examples of adjusting for bias and uncertainties resulting from administrative overestimates, exposures from former employment, and measurement error are examined. Estimates of annual, cumulative, and collective doses are used to examine typical dose distributions and to facilitate dose-response analyses. Work-related medical exposures were also investigated as a potential source of bias. Work-related chest x rays (WRX) can result in a significant fraction of total dose, especially for radiation workers who were more likely to be subjected to routine medical monitoring. Omission of WRX from the total dose category misclassification and may skew the epidemiologic dose-response assessment for cancers induced by the workplace.