Background. Ongoing surveillance of antimicrobial sensitivity patterns of bacteria isolated in bloodstream infections guides empiric antibiotic therapy in neonatal sepsis. Methods. Sensitivity profiles of neonatal bacterial bloodstream infections in a tertiary hospital were reviewed between 01/06/2009 and 30/06/2010 . Results. There were 246 episodes of bloodstream infection in 181 individuals-(14.06 episodes in10.35 patients/1000 patient days or 14.4 episodes in 10.6 babies/1000 live births. The majority were (93.5%) were late onset and most (54.9%) were gram positive. There were 2.28 sepsis-related deaths /1000 patient days or 2.3/1000 live births. Death was significantly associated with gram-negative infections (P < 0.001), multiple gestation (P < 0.001), shock (P = 0.008), NEC (P = 0.002), and shorter duration of hospital stay (P < 0.001). Coagulase-negative staphylococcus was isolated in 19.1%, K. pneumoniae ESBL in 12.1%, and A. baumanni in 10.9%. S. agalactiae predominated in early onset sepsis. Methicillin resistance was present in 86% of CoNS and 69.5% of S. aureus; 46% enterococcal isolates were ampicillin resistant. The majority (65%) of K. pneumoniae isolates were ESBL producers. Ampicillin resistance was present in 96% of E. coli. Conclusions. Penicillin and an aminoglycoside would be suitable empiric therapy for early onset sepsis and meropenem with gentamycin or ceftazidime with amikacin for late onset sepsis.