Background New antibiotics are needed for the treatment of patients with life-threatening carbapenem-resistant Gram-negative infections. We assessed the efficacy and safety of cefiderocol versus best available therapy in adults with serious carbapenem-resistant Gram-negative infections. Methods We did a randomised, open-label, multicentre, parallel-group, pathogen-focused, descriptive, phase 3 study in 95 hospitals in 16 countries in North America, South America, Europe, and Asia. We enrolled patients aged 18 years or older admitted to hospital with nosocomial pneumonia, bloodstream infections or sepsis, or complicated urinary tract infections (UTI), and evidence of a carbapenem-resistant Gram-negative pathogen. Participants were randomly assigned (2:1 by interactive web or voice response system) to receive either a 3-h intravenous infusion of cefiderocol 2 g every 8 h or best available therapy (pre-specified by the investigator before randomisation and comprised of a maximum of three drugs) for 7-14 days. For patients with pneumonia or bloodstream infection or sepsis, cefiderocol treatment could be combined with one adjunctive antibiotic (excluding polymyxins, cephalosporins, and carbapenems). The primary endpoint for patients with nosocomial pneumonia or bloodstream infection or sepsis was clinical cure at test of cure (7 days [plus or minus 2] after the end of treatment) in the carbapenem-resistant microbiological intention-to-treat population (ITT; ie, patients with a confirmed carbapenem-resistant Gram-negative pathogen receiving at least one dose of study drug). For patients with complicated UTI, the primary endpoint was microbiological eradication at test of cure in the carbapenem-resistant microbiological ITT population. Safety was evaluated in the safety population, consisting of all patients who received at least one dose of study drug. Mortality was reported through to the end of study visit (28 days [plus or minus 3] after the end of treatment). Summary statistics, including within-arm 95% CIs calculated using the Clopper-Pearson method, were collected for the primary and safety endpoints. This trial is registered with ClinicalTrials.gov (NCT02714595) and EudraCT (2015-004703-23). Findings Between Sept 7, 2016, and April 22, 2019, we randomly assigned 152 patients to treatment, 101 to cefiderocol, 51 to best available therapy. 150 patients received treatment: 101 cefiderocol (85 [85%] received monotherapy) and 49 best available therapy (30 [61%] received combination therapy). In 118 patients in the carbapenem-resistant microbiological ITT population, the most frequent carbapenem-resistant pathogens were Acinetobacter baumannii (in 54 patients [46%]), Klebsiella pneumoniae (in 39 patients [33%]), and Pseudomonas aeruginosa (in 22 patients [19%]). In the same population, for patients with nosocomial pneumonia, clinical cure was achieved by 20 (50%, 95% CI 33.8-66.2) of 40 patients in the cefiderocol group and ten (53%, 28.9-75.6) of 19 patients in the best available therapy group; for patients with bloodstream infection or sepsis, clinical cure was achieved by ten (43%, 23.2-65.5) of 23 patients in the cefiderocol group and six (43%, 17.7-71.1) of 14 patients in the best available therapy group. For patients with complicated UTIs, microbiological eradication was achieved by nine (53%, 27.8-77.0) of 17 patients in the cefiderocol group and one (20%, 0.5-71.6) of five patients in the best available therapy group. In the safety population, treatment-emergent adverse events were noted for 91% (92 patients of 101) of the cefiderocol group and 96% (47 patients of 49) of the best available therapy group. 34 (34%) of 101 patients receiving cefiderocol and nine (18%) of 49 patients receiving best available therapy died by the end of the study; one of these deaths (in the best available therapy group) was considered to be related to the study drug. Interpretation Cefiderocol had similar clinical and microbiological efficacy to best available therapy in this heterogeneous patient population with infections caused by carbapenem-resistant Gram-negative bacteria. Numerically more deaths occurred in the cefiderocol group, primarily in the patient subset with Acinetobacter spp infections. Collectively, the findings from this study support cefiderocol as an option for the treatment of carbapenem-resistant infections in patients with limited treatment options. Copyright (C) 2020 Elsevier Ltd. All rights reserved.