Dose assessment after radiological disasters is imperative to decrease mortality through rationally directed medical intervention. Our goal was to identify biomarkers capable of qualitative (nonirradiated/irradiated) and/or quantitative (dose) assessment of radiation exposure. Using real-time quantitative PCR, biodosimetry genes were identified in blood samples from cancer patients undergoing total-body irradiation. Time- (5, 12, 23, 48 h) and dose- (0–8 Gy) dependent changes in gene expression were examined in C57BL/6 mice. A training set was used to derive weighted voting classification algorithms (nonirradiated/irradiated) and continuous regression (dose assessment) models that were tested in a separate validation set of mice. Of eight biodosimetry genes identified in cancer patients (ACTA2, BBC3, CCNG1, CDKN1A, GADD45A, MDK, SERPINE1, Tnfrsf10b), expression of BBC3, CCNG1, CDKN1A, SERPINE1 and Tnfrsf10b was significantly (P < 0.05) increased in irradiated mice. CCNG1 and CDKN1A expression segregated irradiated mice from controls with an accuracy, specificity and sensitivity of 96.3, 100.0 and 94.4%, respectively, at 48 h. Multiple linear regression analysis predicted doses for the 0-, 1-, 2-, 4-, 6- and 8-Gy treatment groups as 0.0 ± 0.2, 1.6 ± 1.0, 2.9 ± 1.4, 5.1 ± 2.0, 5.3 ± 0.7 and 10.5 ± 5.6 Gy, respectively. These results suggest that gene expression analysis could be incorporated into biodosimetry protocols for qualitative and quantitative assessment of radiation exposure.