The degree of severity of hematologic acute radiation syndrome (HARS) may vary across the range of radiation doses, such that dose alone may be a less reliable predictor of clinical course. We sought to elucidate the relationship between absorbed dose and risk of clinically relevant HARS in humans. We used the database SEARCH (System for Evaluation and Archiving of Radiation Accidents based on Case Histories), which contains the histories of radiation accident victims. From 153 cases we extracted data on dose estimates using the dicentric assay to measure individual biological dosimetry. The data were analyzed according to the corresponding hematological response categories of clinical significance (H1–4). These categories are derived from the medical treatment protocols for radiation accident victims (METREPOL) and represent the clinical outcome of HARS based on severity categories ranging from 1–4. In addition, the category H0 represents a post-exposure hematological response that is within the normal range for nonexposed individuals. Age at exposure, gender and ethnicity were considered as potential confounders in unconditional cumulative logistic regression analysis. In most cases, victims were Caucasian (82.4%) and male (92.8%), who originated from either the Chernobyl (69.3%) or Goiânia (10.5%) accident, and nearly 60% were aged 20–40 years at time of exposure. All individuals were whole-body exposed (mean 3.8 Gy, stdev ±3.1), and single exposures were predominantly reported (79%). Seventy percent of victims in category H0 were exposed to ≤1 Gy, with rapidly decreasing proportions of H0 seen at doses up to 5 Gy. There were few HARS H4 cases reported at exposed dose of 1–2 Gy, while 82% of H4 cases received doses of >5 Gy. HARS H1–3 cases varied among dose ranges from 1–5 Gy. In summary, single whole-body radiation doses <1 Gy and >5 Gy corresponded in general with H0 and H3–4, respectively, and this was consistent with medical expectations. This underlines the usefulness of dose estimates for HARS prediction. However, whole-body doses between 1–5 Gy poorly corresponded to HARS H1–3. The dose range of 1–5 Gy was of limited value for medical decision-making regarding, e.g., hospitalization for H2–3, but not H1 and treatment decisions that differ between H1–3. Also, there were some H0 cases at high doses and H2–4 cases at low doses, thereby challenging an individual recommendation based solely on dose.