Background The best available treatment against carbapenemase-producing Enterobacteriaceae (CPE) is unknown. The objective of this study was to investigate the effect of appropriate therapy and of appropriate combination therapy on mortality of patients with bloodstream infections (BSIs) due to CPE. Methods In this retrospective cohort study, we included patients with clinically significant monomicrobial BSIs due to CPE from the INCREMENT cohort, recruited from 26 tertiary hospitals in ten countries. Exclusion criteria were missing key data, death sooner than 24 h after the index date, therapy with an active antibiotic for at least 2 days when blood cultures were taken, and subsequent episodes in the same patient. We compared 30 day all-cause mortality between patients receiving appropriate (including an active drug against the blood isolate and started in the first 5 days after infection) or inappropriate therapy, and for patients receiving appropriate therapy, between those receiving active monotherapy (only one active drug) or combination therapy (more than one). We used a propensity score for receiving combination therapy and a validated mortality score (INCREMENT-CPE mortality score) to control for confounders in Cox regression analyses. We stratified analyses of combination therapy according to INCREMENT-CPE mortality score (0–7 [low mortality score] vs 8–15 [high mortality score]). INCREMENT is registered with ClinicalTrials.gov, number NCT01764490. Findings Between Jan 1, 2004, and Dec 31, 2013, 480 patients with BSIs due to CPE were enrolled in the INCREMENT cohort, of whom we included 437 (91%) in this study. 343 (78%) patients received appropriate therapy compared with 94 (22%) who received inappropriate therapy. The most frequent organism was Klebsiella pneumoniae (375 [86%] of 437; 291 [85%] of 343 patients receiving appropriate therapy vs 84 [89%] of 94 receiving inappropriate therapy) and the most frequent carbapenemase was K pneumoniae carbapenemase (329 [75%]; 253 [74%] vs 76 [81%]). Appropriate therapy was associated with lower mortality than was inappropriate therapy (132 [38·5%] of 343 patients died vs 57 [60·6%] of 94; absolute difference 22·1% [95% CI 11·0–33·3]; adjusted hazard ratio [HR] 0·45 [95% CI 0·33–0·62]; p<0·0001). Among those receiving appropriate therapy, 135 (39%) received combination therapy and 208 (61%) received monotherapy. Overall mortality was not different between those receiving combination therapy or monotherapy (47 [35%] of 135 vs 85 [41%] of 208; adjusted HR 1·63 [95% CI 0·67–3·91]; p=0·28). However, combination therapy was associated with lower mortality than was monotherapy in the high-mortality-score stratum (30 [48%] of 63 vs 64 [62%] of 103; adjusted HR 0·56 [0·34–0·91]; p=0·02), but not in the low-mortality-score stratum (17 [24%] of 72 vs 21 [20%] of 105; adjusted odds ratio 1·21 [0·56–2·56]; p=0·62). Interpretation Appropriate therapy was associated with a protective effect on mortality among patients with BSIs due to CPE. Combination therapy was associated with improved survival only in patients with a high mortality score. Patients with BSIs due to CPE should receive active therapy as soon as they are diagnosed, and monotherapy should be considered for those in the low-mortality-score stratum. Funding Spanish Network for Research in Infectious Diseases, European Development Regional Fund, Instituto de Salud Carlos III, and Innovative Medicines Initiative.