A controlled clinical comparison was made of the Isolator (Wampole Laboratories, Cranbury, N.J.) and the Septi-Chek Release bottle (Roche Diagnostics, Nutley, N.J.). From 6,345 blood culture sets fulfilling minimum criteria for volume of blood cultured, 840 strains were isolated, of which only 691 (82%) were considered to be representative of bloodstream infection according to Centers for Disease Control definitions. Statistically significant differences were found between the systems for the following organisms, which were all detected more frequently in the Isolator system: Staphylococcus aureus (P = 0.0001), Alcaligenes xylosoxidans (P = 0.008), Klebsiella pneumoniae (P = 0.05), Salmonella spp. (P = 0.03), and Candida albicans (P = 0.02). The Septi-Chek Release system required a longer period of time than the Isolator system for detection of the following organisms:S. aureus (P = 0.0001), Enterococcus spp. (P = 0.0001), Enterobacter cloacae (P = 0.03), Escherichia coli (P = 0.0001), Klebsiella oxytoca (P = 0.03), K. pneumoniae (P = 0.02), Pseudomonas aeruginosa (P = 0.002), and C. albicans (P = 0.005). There were 430 episodes of bloodstream infections identified in the study; of these episodes, only those due to S. aureus were detected significantly more frequently (P = 0.0001) by the Isolator system than by the Septi-Chek Release system. However, episodes of bloodstream infections due to S. aureus, Staphylococcus epidermidis, Enterococcus spp., and E. coli were detected significantly faster by the Isolator system.