Inadequate stratification of septic shock patients may result in inappropriate treatment allocation in randomized clinical trials, especially regarding anticoagulant. We previously reported that endothelial-derived microparticles are relevant biomarkers of sepsis-induced disseminated intravascular coagulation. In this validation cohort, we assess microparticles as surrogates of cell activation to improve early disseminated intravascular coagulation diagnosis and patient stratification. Prospective observational study in septic shock patients. Four medical ICUs in university hospitals. Two hundred sixty-five patients with septic shock from four ICUs were consecutively enrolled. Disseminated intravascular coagulation was diagnosed according to Japanese Association for Acute Medicine 2006 score. Endothelial- and leukocyte-derived circulating procoagulant microparticles were isolated and quantified by prothrombinase assay at admission, day 3, and day 7. None. Two hundred fifty-nine patients were analyzed. Sixty-one had disseminated intravascular coagulation at admission, and 32 developed disseminated intravascular coagulation during the first 24 hours after admission. Multiple logistic regression model confirmed that endothelial cell-derived microparticles were associated with disseminated intravascular coagulation: CD105-microparticles (odds ratio, 2.13) and CD31-microparticles (odds ratio, 0.65) (p < 0.05). Furthermore, CD11a-microparticles to leukocyte ratio evidenced leukocyte activation (odds ratio, 1.59; p < 0.05). Prediction of disseminated intravascular coagulation was also analyzed after exclusion of patients with disseminated intravascular coagulation at admission. A new multiple logistic regression analysis demonstrated the association of CD105-microparticles (> 0.60 nM eq. PhtdSer; odds ratio, 1.67; p < 0.01), platelets count (≤ 127 g/L; odds ratio, 0.99; p < 0.01), and prothrombin time (≤ 58%; odds ratio, 0.98; p < 0.05) with disseminated intravascular coagulation. A combining score at admission is predictive of the absence of disseminated intravascular coagulation (area under the curve, 72.9%; specificity, 71.2%; sensitivity, 71.0%, with a negative predictive value of 93.1% and a positive predictive value of 31.0%). Procoagulant microparticles from endothelial cells and leukocytes reflect a vascular injury during sepsis-induced disseminated intravascular coagulation that precedes obvious activation of coagulation. A combination of prothrombin time, endothelium-derived CD105-microparticles, and platelet count at admission could predict the absence of disseminated intravascular coagulation and allow a better stratification in future randomized clinical trials.