A large increase in multi-drug-resistant Acinetobacter baumannii, especially carbapenem-resistant strains, occurred during the first two years of the COVID-19 pandemic, posing important challenges in its treatment. Cefiderocol appeared to be a good option for the treatment of Carbapenem-resistant Acinetobacter baumannii (CR-Ab), but to date, the guidelines and evidence available are conflicting. We retrospectively included a group of patients with CR-Ab infections (treated with colistin- or cefiderocol-based regimens) at Padua University Hospital (August 2020-July 2022) and assessed predictors of 30-day mortality, and differences in microbiological and clinical treatment. To evaluate the difference in outcomes, accounting for the imbalance in antibiotic treatment allocation, a propensity score weighting (PSW) approach was adopted. We included 111 patients, 68% males, with a median age of 69 years (IQR: 59-78). The median duration of antibiotic treatment was 13 days (IQR:11-16). In total, 60 (54.1%) and 51 (45.9%) patients received cefiderocol- and colistin-based therapy, respectively. Notably, 53 (47.7%) patients had bloodstream infections, while 58 (52.3%) had pneumonia. Colistin was combined in 96.1%, 80.4%, and 5.8% of cases with tigecycline, meropenem, and fosfomycin, respectively. Cefiderocol was combined in 13.3%, 30%, and 18.3% of cases with fosfomycin, tigecycline, and meropenem, respectively. At the baseline, the two treatment groups significantly differed in age (patients treated with colistin were significantly older), the prevalence of diabetes and obesity (more frequent in the group treated with colistin), length of stay (longer in the group receiving cefiderocol), and type of infection (BSI were more frequent in the group receiving cefiderocol). The proportion of patients who developed acute kidney injury was significantly higher in the colistin group. By using PSW, no statistically significant differences emerged for mortality or clinical and microbiological cure between the two groups. No independent predictors were detected for hospital mortality or clinical cure, while for the length of stay, the only selected predictor was age, with a non-linear effect (p-value 0.025 for non-linearity) on the prolongation of hospital stay of 0.25 days (95% CI 0.10-0.39) at increasing ages (calculated over the IQR). Cefiderocol treatment did not differ in terms of main outcomes and safety profile from colistin-based regimens. More prospective studies with a larger number of patients are required to confirm our results.