Abstract: Objective To retrospectively analyze the prognostic value of platelet count recovery in the early stage post burn for patients with extremely severe burn, so as to evaluate their severity. Methods A study involving 244 adult patients with extremely severe burn admitted to our hospital from January 2006 to December 2015, conforming to the inclusion criteria, was conducted. Data of their demography, injury, transmission, disease change in hospital, and platelet count from post injury day (PID) 1 to 10 were collected. (1) Patients were divided into survival group (n=212) and non-survival group (n=32) according to whether death or not. The dynamic change characteristic of platelet count in patients of two groups from PID 1 to 10 was analyzed and compared. (2) Patients were divided into return to normal group (RN, n=163) and non-return to normal group (NRN, n=81) according to whether platelet count returned to normal within one week post burn. The proportion of patients who received mechanical ventilation and mortality in groups RN and NRN, and length of stay in ICU between patients with platelet count that returned to normal and that did not return to normal in the early stage post burn in survival group were compared. Data were processed with independent samples t test , analysis of variance of repeated measurement, chi-square test, and binomial distribution test. Correlation was analyzed between data of sex, age, weight, total burn area, full-thickness burn area, inhalation injury, length of hospital stay, receiving mechanical ventilation, platelet count recovery condition within one week post burn and death of patients using the univariate and multivariate Cox regression analysis. Receiver operating characteristic (ROC) curve of platelet count on PID 8 was drawn to evaluate predicting value for death of 244 patients. Results (1) The platelet count kept declining from PID 1 to 4, declined to below the normal low limit on PID 2, and reached its lowest level on PID 4 both in survival group and non-survival group. The platelet count both in survival group and non-survival group rose gradually from PID 4 to 10, and returned to normal on PID 8 and 10, respectively. There was no significant difference in platelet count of patients in two groups on PID 1 (t=1.01, P>0.05), while platelet count of patients in non-survival group was obviously lower than that in survival group from PID 2 to 10 (with t values from 2.64 to 7.17, P values below 0.01). The daily increment of platelet count in survival group from PID 4 to 10 was (26±13)×10⁹/L, obviously higher than that in non-survival group [(19±11)×10⁹/L, t=2.76, P<0.01]. (2) The proportion of patients who received mechanical ventilation was obviously lower than that of patients who did not receive mechanical ventilation in group RN (P<0.01), while the proportion of patients who received mechanical ventilation was obviously higher than that of patients who did not receive mechanical ventilation in group NRN (P<0.05). The proportion of patients who received mechanical ventilation in group NRN was obviously higher than that in group RN (χ²=32.93, P<0.01). The mortality of patients in group NRN was obviously higher than that in group RN (χ²=20.99, P<0.01). The length of stay in ICU of patients whose platelet count did not return to normal in the early stage was significantly longer than that of patients whose platelet count returned to normal in the early stage in survival group (t=4.20, P<0.01). (3) Total burn area, receiving mechanical ventilation, and platelet count did not return to normal within one week post burn were independent risk factors for death of patients with extremely severe burn (with hazard ratio respectively 1.073, 16.552, and 2.249, 95% confidence interval respectively 1.033-1.115, 2.147-127.580, and 0.993-5.096, P<0.05 or P<0.01). (4) The area under the ROC curve of platelet count on PID 8 to predict death of 244 patients with extremely severe burn was 0.745 (with 95% confidence interval 0.645-0.845, P<0.01), and 150×10⁹/L was chosen as the optimal threshold value, with sensitivity of 71.4% and specificity of 71.0%. Conclusions Platelet count recovery in the early stage post burn of patients with extremely severe burn was significantly associated with their prognosis and could be used as an important indicator to evaluate the severity of illness.