Abstract: Objective　To explore the prognostic value of platelet count dynamic changes in the 28-day mortality of sepsis. Methods　The consecutive patients with sepsis hospitalized from January 2018 to December 2019 in our hospital were retrospectively studied. Dynamic data of platelet counts (PLT) within 14 days after ICU admission were collected. Progressive thrombocytopenia (PTCP) was identified as PLT decreasing over 20 percent than the baseline (the day 1) on the day 7. The performance of PTCP and sequential organ failure (SOFA) in predicting the 28-day mortality was evaluated via ROC analysis. The Logistic regression model was constructed to investigate the risk factors for the 28-day mortality of sepsis. The survival status of patients with and without PTCP were compared through Kaplan-Meier survival analysis. Results　A total of 148 patients were included in this study, 30 in the non-survival and others in the survival group. The 28-day mortality was 20.3%. In comparison with survival group, the SOFA score, hD-dimer, TCP and PTCP were significantly higher in non-survival group (P＜0.01). Five days after admission, PLT showed a tendency of progressive decline in non-survival group. The numbers of PLT were significantly higher on the day 7, 10, and 14 in survival group than those of non-survival group (P＜0.01). The area under ROC curves for PTCP, SOFA and their combination in predicting the 28-day mortality of sepsis were 0.683, 0.691 and 0.778, respectively. Logistic regression analysis indicated that PTCP was an independent risk factor for the 28-day mortality of sepsis (P＜0.01). Kaplan Meier survival analysis showed that the 28 day survival rate was significantly reduced in patients with PTCP (P＜0.01). Conclusion　PTCP is an independent risk factor for the 28-day mortality of sepsis. PTCP has a certain predictive value for the 28-day mortality of sepsis, and combined with baseline SOFA score can improve predictive efficiency.