PURPOSE: The aim of this study was to assess associations between consecutive thrombelastography (TEG) profiles and standard coagulation tests and disease severity and mortality in patients with severe sepsis. MATERIALS AND METHODS: This was a prospective observational study of adults with severe sepsis admitted to the intensive care unit (ICU). Clinical scores/variables, infection, TEG, biochemistry, therapy, and overall mortality were recorded. RESULTS: Fifty patients (60% men, median age 62 years, 28-day mortality 24%) were included. At admission, 22%, 48%, and 30% had a hypocoagulable, normocoagulable, and hypercoagulable TEG clot strength (maximum amplitude [MA]), respectively. Hypocoagulable patients had higher Sequential Organ Failure Assessment and disseminated intravascular coagulation scores compared with hypercoagulable patients and higher 28-day mortality compared with normocoagulable patients (all P <.05). Most patients (73%-91%) displayed a TEG MA comparable with the admission profile during the initial 4 ICU days or until death/discharge. Patients progressing to hypocoagulable MA had a high early mortality (80%) and hypocoagulable MA independently predicted 28-day mortality (adjusted odds ratio, 4.29 [95% confidence interval, 1.35-13.65], P = .014). In hypocoagulable and hypercoagulable patients, only fibrinogen (P = .041 and P <.001, respectively) contributed independently to clot strength, whereas both platelets (P <.001) and fibrinogen (P <.001) contributed independently to clot strength in normocoagulable patients. CONCLUSIONS: The ICU admission TEG MA remained constant for several days in patients with severe sepsis and hypocoagulable MA independently predicted 28-day mortality.