Radiation exposure from dental and medical radiography is inherently difficult to estimate. Because no single resource is available for this type of data, gathering the information needed to estimate exposure in epidemiologic studies is a labor-intensive, imprecise process. For a case-control study of adult-onset acute myelogenous leukemia in Los Angeles County and radiography, a database was created of estimates of dose to the red bone marrow for each radiographic procedure reported by subjects in interviews and recorded in subjects' medical records. Resources used included the medical literature as well as personal communications with radiology experts. Dose estimates for each procedure based on this database are reported. Methods and complications are contrasted with a past effort to estimate dose from dental radiography for a case-control study of parotid gland tumors. Among the more difficult aspects of medical radiography dose estimation are the wide variety of examinations performed, the continually-changing environment of diagnostic imaging, and the number of variables that contribute to the delivered dose to an individual from a specific imaging procedure.