There is minimal data evaluating the safety of antibiotic de-escalation in patients with acute myeloid leukemia (AML) with fever and ongoing neutropenia. Therefore, this study evaluated antibiotic prescribing, infection-related outcomes, and patient outcomes of an antibiotic de-escalation initiative. This pre-post quasiexperimental study included adult patients with AML hospitalized with febrile neutropenia. An antibiotic de-escalation guideline was implemented in January 2017, which promoted de-escalation or discontinuation of intravenous antipseudomonal β-lactams. The primary outcome assessment was the incidence of bacterial infection in a historical control group before guideline implementation compared with an intervention group after guideline implementation. A total of 93 patients were included. Antibiotic de-escalation occurred more frequently in the intervention group (71.7% vs 7.5%; P<.001), which resulted in fewer days of therapy for intravenous antipseudomonal β-lactams (14 vs 25 days; P<.001). Thirty-day all-cause mortality and length of hospitalization were not different between groups. However, the intervention group had significantly fewer episodes of Clostridioides difficile colitis (5.7% vs 27.5%; P=.007). Implementation of an antibiotic de-escalation guideline resulted in decreased use of intravenous antipseudomonal β-lactams and fewer episodes of C difficile colitis, without adversely impacting patient outcomes. Additional studies are needed, preferably in the form of randomized controlled trials, to confirm these results.