Dosimetry is an important tool for triage and treatment planning following any radiation exposure accident, and biological dosimetry, which estimates exposure dose using a biological parameter, is a practical means of determining the specific dose an individual receives. The cytokinesis-blocked micronucleus assay (CBMN) is an established biodosimetric tool to measure chromosomal damage in mitogen-stimulated human lymphocytes. The CBMN method is especially valuable for biodosimetry in triage situations thanks to simplicity in scoring and adaptability to high-throughput automated sample processing systems. While this technique produces dose-response data which fit very well to a linear-quadratic model for exposures to low linear energy transfer (LET) radiation and for doses up for 5 Gy, limitations to the accuracy of this method arise at larger doses. Accuracy at higher doses is limited by the number of cells reaching mitosis. Whereas it would be expected that the yield of micronuclei increases with the dose, in many experiments it has been shown to actually decrease when normalized over the total number of cells. This variation from a monotonically increasing dose response poses a limitation for retrospective dose reconstruction. In this study we modified the standard CBMN assay to increase its accuracy following exposures to higher doses of photons or a mixed neutron-photon beam. The assay is modified either through inhibitions of the G2/M and spindle checkpoints with the addition of caffeine and/or ZM447439 (an Aurora kinase inhibitor), respectively to the blood cultures at select times during the assay. Our results showed that caffeine addition improved assay performance for photon up to 10 Gy. This was achieved by extending the assay time from the typical 70 h to just 74 h. Compared to micronuclei yields without inhibitors, addition of caffeine and ZM447439 resulted in improved accuracy in the detection of micronuclei yields up to 10 Gy from photons and 4 Gy of mixed neutrons-photons. When the dose-effect curves were fitted to take into account the turnover phenomenon observed at higher doses, best fitting was achieved when the combination of both inhibitors was used. These techniques permit reliable dose reconstruction after high doses of radiation with a method that can be adapted to high-throughput automated sample processing systems.