The optimal duration of antibiotic treatment for critically ill patients with bloodstream infections is unknown. There is evidence in less severe infections that shorter durations of antibiotic treatment are as equally effective as longer durations. There is considerable variation in the length of antibiotic treatment doctors prescribe for patients with bloodstream infection, but longer durations contribute to antibiotic resistance, complications such as secondary infections, and increased healthcare costs. This international, multi-center, randomised, controlled trial in New Zealand, Canada and Australia involving 3,600 patients in 40 intensive care units will determine whether shorter duration antibiotic treatment (7 days) for patients with bloodstream infection is as effective as longer treatment (14 days) and associated with less total antimicrobial use, and fewer secondary bowel infections, adverse events, and development of antibiotic-resistant organisms. The cost-effectiveness of total treatment costs under the two alternatives will also be studied.