The glycophorin A (GPA)-based human in vivo somatic cell mutation assay, because of it's demonstrated long-term biological memory of past human exposures to ionizing radiation, together with it's relatively low cost and high sample thruput, is unique among presently available human biomarker assays for practical use as a retrospective radiation biodosimeter in epidemiological investigations of large human populations in environmental or occupational exposure settings. This study was undertaken to further validate the response of the assay as a biodosimeter of low dose/low dose rate radiation exposure in a longitudinal study of patients receiving 131I therapy for thyroid disease. These patients are excellent subjects for this investigation as they receive clinically well characterized doses ranging typically from 10 to 100 cSv of whole body bone marrow exposure to ionizing radiation resulting from the radioactive decay of administered 131I. The study design entailed the collection and analysis of multiple peripheral blood samples from patients drawn prior to, during, and following therapy to follow the induction, accumulation, and persistence of radiation-induced somatic mutation at the GP A locus in bone marrow stem cells. These mutations in nucleated bone marrow progenitor cells give rise to erythrocytes in the peripheral circulation expressing a GPA allele-loss variant phenotype. These variants are directly enumerated in the assay using immunolabeling with GPA allele-specific monoclonal antibodies and flow cytometry. Based on GP A assay results obtained in populations with high/high dose rate radiation exposures, this study was designed to primarily investigate the radiation-dose response of the assay over a range of doses that surround the extrapolated doubling dose over background response of the assay of approximately 30 cGy. The longitudinal design of the study, applied to patients receiving relatively low doses of 131I, permitted a critical examination of the practical ultimate sensitivity of the assay by comparing GPA variant cell frequencies (Vi) in post-therapy samples to those observed in pre-therapy samples within individual patients. The results of the study provide critical information to assess the power of the assay to demonstrate, or to estimate the upper limits of, radiation exposures in population surveys of environmentally-exposed populations or occupationally-exposed workers. To date, 17 thyroid cancer patients, treated with ablative l31I therapy, have been enrolled in the study and complete GP A assay data has been obtained and analyzed for eight patients who provided a time-series of blood samples collected prior to, and 14, 30, 50, 90, 120, 180, and 365 days after administration of 97.5 - 203 (mean 126) mCi of 131I. A consistent short-term induction and partial long-term persistence of GPA allele-loss Vf was observed suggestive of radiation-induced GPA locus somatic mutation in both committed erythroid lineage and bone marrow stem cell populations resulting from a calculated mean marrow dose of 34 cSv. The persistent mean increase of GPA Vf, measured one year after exposure, was 1.9 per million cells compared to 8.5 predicted from the dose-response relationship derived from the previous high dose/high dose rate studies, suggesting a DDREF of approximately 4.5 associated with this exposure. This result, consistent with assay data obtained previously in populations of Chernobyl cleanup workers with protracted radiation exposures, demonstrates that, like other experimental and biodosimetric systems, the GPA assay yields a substantially reduced biological response to low dose/low dose rate exposures.