In sepsis-associated coagulopathies and disseminated intravascular coagulation, relative platelet reductions may reflect coagulopathy severity. However, limited evidence supports their clinical significance and most sepsis-associated coagulopathy criteria focus on the absolute platelet counts. To estimate the impact of relative platelet reductions and absolute platelet counts on sepsis outcomes. A multicenter retrospective observational study was performed using the eICU Collaborative Research Database, comprising 335 intensive care units (ICUs) in the United States. Patients with sepsis and an ICU stay > 2 days were included. Estimated effects of relative platelet reductions and absolute platelet counts on mortality and coagulopathy-related complications were evaluated. Overall, 26,176 patients were included. Multivariate mixed-effect logistic regression analysis revealed marked in-hospital mortality risk with larger platelet reductions between days one and two, independent from the resultant absolute platelet counts. The adjusted odds ratio (OR) [95% confidence intervals (CI)] for in-hospital mortality was 1.28[1.23-1.32], 1.86[1.75-1.97], 2.99[2.66-3.36], and 6.05[4.40-8.31] for 20-40%, 40-60%, 60-80%, and > 80% reductions, respectively, when compared with a < 20% decrease in platelets (P < 0.001 for each). In the multivariate logistic regression analysis, platelet reductions ≥ 11% and platelet counts ≤ 100,000/μL on day 2 were associated with high coagulopathy-related complications (OR [95%CI], 2.03 and 1.18; P < 0.001 and P < 0.001), while only platelet reduction was associated with thromboembolic complications (OR [95%CI], 1.43 [1.03-1.98], P < 0.001). The magnitude of platelet reductions represent mortality risk and provides a better signature of coagulopathies in sepsis; therefore, it is a plausible criterion for sepsis-associated coagulopathies.