Two hundred twenty-six patients with bacteremia were prospectively enrolled in a randomized trial that was performed to determine the clinical impact of the receipt of in vitro microbiological data by the physician soon after organism detection as opposed to having the physicians wait until similar data were available by routine methods. Identification and antibiotic susceptibility patterns of 110 isolates were determined by direct inoculation of the Vitek AutoMicrobic system (Vitek Systems, Inc., Hazelwood, Mo.) with a sample from a positive blood culture vial. One hundred sixteen isolates were processed by routine methods. Microbiological results were available within an average of 8.8 h by the direct method versus an average of 48 h by the routine method. In both groups an infectious disease fellow used the information to make therapeutic recommendations to the responsible physician. When compared with that provided by the routine method, the information provided by the direct method was significantly more likely to result in an initiation of antibiotic therapy, a change to more effective therapy, or a change to less expensive therapy. Recommendations were significantly more likely to be followed in patients whose isolates were processed by the direct method versus the routine method. A projected savings of $158 per patient was estimated for the patients who were changed to less expensive therapy or in whom antibiotics were discontinued because results were available sooner. These cost savings, coupled with changes in therapy made for reasons of efficacy, support the usefulness of the earlier reporting of the identity and antibiotic susceptibility patterns of bacterial blood culture isolates.