Abstract An optimum duration of antibiotic therapy would eradicate infection whilst minimising adverse drug reactions, resistance and costs. However, there is a paucity of evidence guiding the duration of therapy for bloodstream infections. Canadian infectious diseases (ID) and critical care specialists were surveyed regarding their recommended antibiotic treatment durations for five common bacteraemic syndromes. A descriptive analysis was performed to assess: (i) practice heterogeneity; (ii) equipoise for a trial of shorter versus longer therapy; and (iii) the influence of pathogen and host characteristics on recommendations. In total, 172 clinicians responded to the survey (60% ID, 39% critical care and 1% combined specialists). For each syndrome, the most common recommendation was 14 days, yet the majority of respondents recommended durations of ≤10 days. Median durations were similar for each syndrome: central vascular catheter-related bloodstream infection, 10 ± 3.6 days; bacteraemic pneumonia, 10 ± 2.8 days; bacteraemic urinary tract infection, 10 ± 3.8 days; bacteraemic intra-abdominal infection, 10 ± 4.1 days; and bacteraemic skin and soft-tissue infection, 14 ± 3.2 days. Respondents recommended the longest durations for Staphylococcus aureus and the shortest durations for coagulase-negative staphylococci. Most respondents would not modify duration based on host characteristics or measures of clinical response. ID physicians recommended longer durations than critical care physicians for all five syndromes, but the majority of both specialist groups would enrol patients in a trial of shorter (7 day) versus longer (14 day) antibiotic therapy. In conclusion, significant practice variation exists amongst clinicians’ recommended durations of antibiotic treatment for bacteraemia. There is equipoise for a randomised trial comparing shorter versus longer courses of antibiotics for most bacteraemic syndromes and pathogens.