Acute febrile illnesses (AFIs) with thrombocytopenia, such as dengue, leptospirosis, and scrub typhus, are often associated with bleeding and present challenges in transfusion therapy during outbreaks. Conventional bleeding predictors, which involve platelet counts, do not assess clot strength or fibrinolysis defects and may lead to inappropriate or delayed transfusions. To predict bleeding, we aimed to create a scoring system using point-of-care thromboelastography (TEG) and other conventional coagulation parameters. In this prospective cohort study, patients with AFI and platelet counts <100,000/cu mm who presented to the emergency department were recruited, and samples were collected for conventional coagulation tests and TEG analysis at admission. Patients were monitored for 48 hours for bleeding events, blood product transfusions, and mortality within 7 days. Receiver operating characteristic curves were plotted for significant predictors identified in univariate analysis, and a scoring system was generated using stepwise logistic regression. Sixty-six patients were recruited, including 38 with dengue and 11 with leptospirosis. Nineteen patients experienced a bleeding event within 48 hours. Dyspnea and elevated aspartate transaminase levels were more prevalent among those who bled. Thromboelastography parameters (reaction time/maximum amplitude [R/MA]), prothrombin time, international normalized ratio (INR), and baseline platelet count were significant predictors of bleeding. A scoring system using R/MA, INR, and platelet count achieved an overall predictive accuracy of 83.3% for 48-hour bleeding, with a score ≥3 demonstrating a sensitivity of 68% and specificity of 77%. Thromboelastography in patients with dengue showed more significant coagulation abnormalities than in those with leptospirosis. Validation in larger, more homogeneous populations is necessary.