This study aimed to identify potential knowledge-performance gaps in antibiotic prescribing for bacterial isolates in the Intensive Care Unit (ICU) in order to guide the development of interventions such as antibiotic policies, decision support, and improved systems for communication between the laboratory and the bedside. A prospective observational cohort study of all patients admitted to a mixed medical/surgical ICU was undertaken over a six-month period in an Australian adult tertiary hospital. From a cohort of 524 patients, 108 had 303 isolates that were eligible for inclusion. Overall, 14.3% and 30.8% of sterile and non-sterile isolates respectively were associated with inadequate initial antibiotic therapy after identification of the bacteria. After sensitivity results were available inadequate directed therapy was observed in 4.0% and 21.3% of sterile and non-sterile isolates respectively. Problems were most commonly associated with isolates of Pseudomonas spp., Stenotrophomonas spp., Acinetobacter spp., S. aureus, enterococci and group III Enterobacteriaceae. Inadequate antibiotic therapy was found to be independently associated with prolonged length of ICU stay. Narrower spectrum antibiotic therapy was potentially available for 30% of isolates after sensitivity results were known. We conclude that there is scope to improve antibiotic prescribing in the ICU by providing clinicians with access to information regarding local susceptibility patterns and intrinsic resistance of bacteria, and spectra of antibiotic cover. Timely notification of laboratory results at the point of care may also facilitate improved prescribing performance.