Retrospective study was conducted in surgical intensive care unit (ICU) in Clinical Hospital Center Zagreb in 2005. The aim of study was to create guidelines for empirical antibiotic therapy of sepsis in ICU for unknown causative agent based on antimicrobial susceptibility of causative bacteria. Thirty-two patients with severe sepsis were included in study and from medical records their clinical and microbiological data were analyzed. Antimicrobial susceptibility of the strains isolated from the blood-culture was tested by disk diffusion method according to CLSI (Clinical Laboratory Standard Institution). We used APACHE II score to predict the severity of illness. Mann-Whitney test and chi2 test were used to test statistical significance difference between results. Acinetobacter baumannii and Pseudomonas aeruginosa were the predominant causative agent. Acinetobacter baumannii was displaying excellent susceptibility to ampicillin+sulbactam and carbapenems, whereas Pseudomonas aeruginosa was showed good susceptibility on ceftazidim and carbapenems. Methicillin-resistant Staphylococcus aureus (MRSA), third predominant causative agent exhibiting good susceptibility to vancomycin and linezolide. The recommended therapy is empirical antibiotic therapy and should cover all important pathogens.