Abstract Background Rapid identification (ID) and antibiotic susceptibility testing (AST) of the causative micro-organism of bloodstream infections result in earlier targeting of antibiotic therapy. In order to obtain results of ID and AST up to 24 hours earlier, we evaluated the accuracy of direct inoculation of the Phoenix system from positive blood cultures (BACTEC) by using Serum Separator Tubes to harvest bacteria from positive blood cultures. Results were compared to those of standard Phoenix procedure. Discrepancies between the two methods were resolved by using the API system, E-test or microbroth dilution. Results ID with the direct method was correct for 95.2% of all tested Enterobacteriaceae (n = 42) and 71.4% of Pseudomonas aeruginosa strains (n = 7). AST with the direct method showed a categorical agreement for Gram-negative rods (GNR) of 99.0%, with 0.7% minor errors, 0.3% very major errors and no major errors. All antibiotics showed an agreement of >95%. The direct method for AST of Staphylococcus (n = 81) and Enterococcus (n = 3) species showed a categorical agreement of 95.4%, with a minor error rate of 1.1%, a major error rate of 3.1% and a very major error rate of 0.4%. All antibiotics showed an agreement of >90%, except for trimethoprim-sulfamethoxazole and erythromycin. Conclusions Inoculation of Phoenix panels directly from positive blood cultures can be used to report reliable results of AST of GNR a day earlier, as well as ID-results of Enterobacteriaceae. For Staphylococcus and Enterococcus species, results of AST can also be reported a day earlier for all antibiotics, except for erythromycin and trimethoprim-sulfamethoxazole.