Changes in hematopoiesis that occur in humans after a burn injury may have important effects on morbidity and mortality. In patients with a variety of severe diseases, the presence of erythroblasts in peripheral blood is known to be indicative of a poor prognosis. However, the prognostic significance of erythroblasts in peripheral blood of burn patients has not yet been estimated. This study included 464 consecutive burn patients, of whom 81 did not survive their injuries (17.5 percent). Together with erythroblasts in blood, data on age, sex, total burn surface area, third-degree burn, inhalation trauma, white blood cell count, C-reactive protein, and hemoglobin were studied. The mortality rate of patients with erythroblasts in peripheral blood (n = 53) amounted to 56.6 percent (n = 30; total burn surface area, 39 percent), which is significantly higher (p < 0.001) than the mortality rate of patients without erythroblasts (12.4 percent, n = 51; total burn surface area, 18.69 percent). None of the 10 patients with more than 1000 erythroblasts x 10/liter survived. The detection of erythroblasts in the peripheral blood of burn patients is highly predictive of death, with the odds ratio after adjustment for the other known prognostic factors being 8.3 (95 percent confidence interval, 4.5 to 15.3). Erythroblasts were detected for the first time on average 10 +/- 4 days (median, 6 days) after admission and 13 +/- 6 days (median, 7 days) before death. Detection of erythroblasts in burn patients is of high prognostic power with regard to in-hospital mortality, providing physicians with a strong prognostic method with which to identify seriously threatened patients. It seems attractive to think about an incorporation of erythroblasts into further refinements of burn scores.