The lysis-centrifugation technique (ISOLATOR; E. I. du Pont de Nemours & Co., Wilmington, Del.) and the radiometric blood culture technique (BACTEC; Johnston Laboratories, Inc., Cockeysville, Md.) were compared on 1,000 blood cultures. A total of 16 ml of blood was distributed: 8 ml into an ISOLATOR 7.5 microbial tube and 4 ml each into BACTEC 7C and 8B bottles. The concentrate from the ISOLATOR tubes was inoculated under a laminar-flow hood onto two sheep blood agar plates (one incubated in CO2 and one incubated anaerobically), one chocolate agar plate, and one brain heart infusion agar plate. Of 91 blood specimens obtained that yielded clinically significant organisms, 52 were positive by both systems, 27 were positive by the ISOLATOR system only, and 12 were positive by the BACTEC system only. From the positive blood specimens, 97 clinically significant organisms were isolated: 57 by both systems, 27 by the ISOLATOR system only, and 13 by the BACTEC system only. Of the 57 organisms detected by both systems, 28 were detected simultaneously, 13 were detected earlier by the ISOLATOR system, and 16 were detected earlier by the BACTEC system. Isolated colonies were obtained earlier by the ISOLATOR system in 40 cases and by the BACTEC system in 5 cases. Organisms determined to be contaminants by thorough chart review were isolated from 138 ISOLATOR tubes. In 98 instances, these were represented by one colony of Staphylococcus epidermidis, alpha-hemolytic streptococci, or diphtheroids. The ability to determine CFU per milliliter with the ISOLATOR system did not help differentiate clinically significant organisms from contaminants.