Background: Early recognition and treatment of sepsis would improve patients` outcome. But it is difficult to distinguish between sepsis and non-infectious conditions in the acute phase of clinical deterioration. We studied serum level of procalcitonin (PCT) as a method to diagnose and to evaluate sepsis. Methods: Between 1 March 2009 and 30 September 2009, 178 patients had their serum PCT tested during their clinical deterioration in the medical intensive care unit. These laboratories were evaluated, on a retrospective basis. We classified their clinical status as non-infection, local infection, sepsis, severe sepsis, and septic shock. Then, we compared their clinical status with level of PCT. Results: The number of clinical status is as follows: 18 non-infection, 33 local infection, 39 sepsis, 26 severe sepsis, and 62 septic shock patients. PCT level of non-septic group (non-infection and local infection) and septic group (sepsis, severe sepsis, septic shock) was 0.36±0.57 ng/mL and 18.09±36.53 ng/mL (p＜0.001), respectively. Area under the curve for diagnosis of sepsis using cut-off value of PCT ＞0.5 ng/mL was 0.841 (p＜0.001). Level of PCT as clinical status was statistically different between severe sepsis and septic shock (*severe sepsis; 4.53±6.15 ng/mL, *septic shock 34.26±47.10 ng/mL, *p＜0.001). Conclusion: Level of PCT at clinical deterioration showed diagnostic power for septic condition. The level of PCT was statistically different between severe sepsis and septic shock.