Aim: To test a modified protocol designed to detect Gram-negative bacilli (GNB) resistant to oximinocephalosporins and carbapenems from positive blood cultures. Design: This is a prospective, cohort study of consecutive patients. Setting: A cardiovascular and University referral hospital. Patients: Patients hospitalized in a third level hospital with bacteraemia. Main variables of interest: We developed a modified protocol using HB&L® system to detect MDRP. We then attempted to determine accuracy, concordance and reduction of identification time of this novel method in a reference hospital. Descriptive statistics and logistical regressions were used. Results: Ninety-six patients with BSI were included in the study. A total of 161 positive blood cultures were analysed. Escherichia coli (50%, 81/161) was the most frequently identified pathogen followed by Klebsiella pneumoniae (15%, 24/161) and Pseudomonas aeruginosa (8%, 13/161). Thirty-two percent of isolations had usual resistance patters. However, in 34/161 (21%) of identified pathogens were producer of carbapenemasases and 21/161 (13%) of extended-spectrum β-lactamases. Concordance among our HB&L®modified protocol and traditional method was 99% (159/161). Finally, identification times were significantly shorter using our HB&L® modified protocol than traditional methods (Median [IQR]; 19 hours [18, 22] Vs 61 hours [60, 64], p<0.001). Conclusions: Here we provided novel evidence that using our HB&L® modified protocol is an effective strategy to reduce the time to MDRP detection/identification; with a great concordance rate when compared to the gold standard. Further studies are needed to confirm these findings and to determine whether this method may improve clinical outcomes.