Differences in cancer occurrence within a racial/ethnic group may be influenced by migrant status or level of acculturation, which can indicate variations in environmental exposures. When immigration and acculturation data on individual patients are not available, birthplace may serve as a proxy. As part of a larger study intended to assess the utility of the birthplace variable in the Greater Bay Area Cancer Registry (GBACR), part of the Surveillance, Epidemiology, and End Results (SEER) program, we measured its completeness, and determined whether missing birthplace is associated with patient and hospital reporting source characteristics. Subjects were persons diagnosed with cancer within the GBACR surveillance area during the period 1988-1996. Of the 204,553 subjects, 67% had birthplace recorded in the registry. Deceased persons were 10 times more likely than living persons to have data on birthplace. Racial/ethnic groups which included more foreign-born persons, such as Southeast Asians, tended to have more complete birthplace information than did groups with fewer foreign-born, such as Japanese and Hispanics. The effects of patient and reporting source characteristics on birthplace completeness differed across racial/ethnic groups. These data indicate that completeness of the birthplace variable in the GBACR is biased, and that investigators considering birthplace in analyses of SEER data should consider these biases. For birthplace data to be useful, completeness needs to be improved at the level of the diagnosing facility.