The research for high-throughput diagnostic tests for victims of radio/nuclear incidents remains ongoing. In this context, we have previously identified candidate genes that predict risk of late-occurring hematologic acute radiation syndrome (HARS) in a baboon model. The goal of the current study was to validate these genes after radiation exposure in humans. We also examined ex vivo relative to in vivo measurements in both species and describe dose-response relationships. Eighteen baboons were irradiated in vivo to simulate different patterns of partial- or total-body irradiation (TBI), corresponding to an equivalent dose of 2.5 or 5 Sv. Human in vivo blood samples were obtained from patients exposed to different dose ranges: diagnostic computerized tomography (CT; 0.004–0.018 Sv); radiotherapy for prostate cancer (0.25–0.3 Sv); and TBI of leukemia patients (2 × 1.5 or 2 × 2 Sv, five patients each). Peripheral whole blood of another five baboons and human samples from five healthy donors were cultivated ex vivo and irradiated with 0–4 Sv. RNA was isolated pairwise before and 24 h after irradiation and converted into cDNA. Gene expression of six promising candidate genes found previously by us in a baboon model (WNT3, POU2AF1, CCR7, ARG2, CD177, WLS), as well as three genes commonly used in ex vivo whole blood experiments (FDXR, PCNA, DDB2) was measured using qRT-PCR. We confirmed the six baboon candidate genes in leukemia patients. However, expression for the candidate gene FDXR showed an inverse relationship, as it was downregulated in baboons and upregulated in human samples. Comparisons among the in vivo and ex vivo experiments revealed the same pattern in both species and indicated peripheral blood cells to represent the radiation-responsive targets causing WNT3 and POU2AF1 gene expression changes. CCR7, ARG2, CD177 and WLS appeared to be altered due to radiation-responsive targets other than the whole blood cells. Linear dose-response relationships of FDXR, WNT3 and POU2AF1 using human ex vivo samples corresponded with human in vivo samples, suggesting that ex vivo models for in vivo dose estimates can be used over a wide dose range (0.001–5 Sv for POU2AF1). In summary, we validated six baboon candidate genes in humans, but the FDXR measurements underscored the importance of independent assessments even when candidates from animal models have striking gene sequence homology to humans. Since whole blood cells represented the same radiation-responsive targets for FDXR, WNT3 and POU2AF1 gene expression changes, ex vivo cell culture models can be utilized for in vivo dose estimates over a dose range covering up to 3.5 log scales. These findings might be a step forward in the development of a gene expression-based high-throughput diagnostic test for populations involved in large-scale radio/nuclear incidents.