Objectives: We studied the performance of aminoglycosides in treating bloodstream infections (BSIs) of urinary source caused by ESBL-producing Enterobacteriaceae (ESBL-EB). Methods: In a retrospective study of 193 patients with a clinical diagnosis of urinary tract infection, pyelonephritis or urosepsis and blood and urine cultures positive for ESBL-EB, patients were grouped according to whether they were treated with an aminoglycoside, a carbapenem or piperacillin/tazobactam. Multivariate analysis was used to define risk factors for mortality with inverse probability of treatment weighting used to minimize confounding. The primary efficacy outcome was 30 day mortality. The primary safety outcome was acute kidney injury (AKI) at 14 days. Results: Mean age was 79.3 years. Dementia, chronic kidney disease and the presence of a urinary catheter were common. Thirty-two (16.6%) patients died and risk factors for mortality included age, high Charlson score, presentation with severe sepsis/septic shock and infection with bacteria other than Escherichia coli. Aminoglycosides were non-inferior compared with other antibiotics regarding 30 day mortality [13.0% versus 21.2%, respectively; adjusted risk difference=10.29% (-0.82% to 21.41%)], but did not reach non-inferiority for bacteriuria recurrence [48.9% versus 44.7%, respectively; adjusted risk difference=-8.72% (-30.87% to 13.43%)]. AKI developed at a similar rate in both treatment groups: 12.0% versus 10.6%, respectively [OR=1.14 (0.46-2.81)]. Aminoglycosides were more efficacious in E. coli infections compared with other ESBL-EB. Conclusions: We demonstrated the efficacy and safety of aminoglycosides in treating BSI of urinary source caused by ESBL-EB. This carbapenem-sparing approach can assist in avoiding excessive carbapenem use without compromising outcomes.