The utility of early-phase (≤5 days) radiation-induced clinical signs and symptoms (e.g., vomiting, diarrhea, erythema and changes in blood cell counts) was examined for the prediction of later occurring acute radiation syndrome (ARS) severity and the development of medical management strategies. Medical treatment protocols for radiation accident victims (METREPOL) was used to grade ARS severities, which were assigned response categories (RCs). Data on individuals (n = 191) with mild (RC1, n = 45), moderate (RC2, n = 19), severe (RC3, n = 20) and fatal (RC4, n = 18) ARS, as well as nonexposed individuals (RC0, n = 89) were generated using either METREPOL (n = 167) or the system for evaluation and archiving of radiation accidents based on case histories (SEARCH) database (n = 24), the latter comprised of real-case descriptions. These data were converted into tables reflecting clinical signs and symptoms, and submitted to eight teams representing five participating countries. The teams were comprised of medical doctors, biologists and pharmacists with subject matter expertise. The tables comprised cumulated clinical data from day 1-3 and day 1-5 postirradiation. While it would have reflected a more realistic scenario to provide the data to the teams over the course of a 3- or 5-day period, the logistics of doing so proved too challenging. In addition, the team members participating in this exercise chose to receive the cumulated reports of day 1-3 and 1-5. The teams were tasked with predicting ARS incidence, ARS severity and the requirement for hospitalization for multiple cases, as well as providing the certainty of their diagnosis. Five of the teams also performed dose estimates. The teams did not employ harmonized methodologies, and the expertise among the members varied, as did the tools used and the means of analyzing the clinical data. The earliest report time was 3 h after the tables were sent to the team members. The majority of cases developing ARS (89.6% ± 3.3 SD) and requiring hospitalization (88.8% ± 4.6 SD) were correctly identified by all teams. Determination of ARS severity was particularly challenging for RC2-3, which was systematically overestimated. However, RC4 was correctly predicted at 94-100% by all teams. RC0 and RC1 ARS severities were more difficult to discriminate. When reported RCs (0-1 and 3-4) were merged, on average 89.6% (±3.3 SD) of all cases could be correctly classified. Comparisons on frequency distributions revealed no statistically significant differences among the following: 1. reported ARS from different teams (P > 0.2); 2. cases generated based on METREPOL or SEARCH (P > 0.5); or 3. results reported at day 3 and 5 postirradiation (P > 0.1). Dose estimates of all teams increased significantly along with ARS severity (P < 0.0001) as well as with dose estimates generated from dicentric chromosomal-aberration measurements available for SEARCH cases (P < 0.0001). In summary, early-phase radiation-induced clinical signs and symptoms proved to be useful for rapid and accurate assessment, with minor limitations, toward predicting life-threatening ARS severity and developing treatment management strategies.