Worker dose estimatesare used in epidemiologicstudiesof nuclear workers. A major objectiveof these studies is to provide a direct assessment. of the carcinogenicrisk of exposure to ionizingradiationat low doses and dose rates. If dose estimatesused in analyses of worker data are biased, • then risk estimatesexpressedper unit of dose will also be biased. In addition,random error in dose estimatesmay lead to underestimationof risk coefficientsand can also distort dose-responseanalyses. Analyses of data from nuclearworker studies, includingHanford, have typicallynot been ' adjusted for biases and uncertaintiesin dose estimates in part because of the lack of adequate informationon the nature and magnitude of these biases and uncertainties. 10%, and bone marrow dose was overestimatedby about 50%. However, these overall bias factors are applicableonly to the "average"worker. Also, they reflect subjectivejudgments and thus are subjectto uncertainty.