The bioMerieux VITAL automated blood culture system measures a decrease in fluorescence to detect the presence of microorganisms in blood. To assess the performance of VITAL with AER aerobic medium versus that of the nonradiometric BACTEC NR-660 PEDS PLUS medium for the detection of sepsis in children, a total of 12,146 blood specimens were collected at three university medical centers and inoculated into AER and PEDS PLUS bottles that were weighed before and after filling. The sample volumes were considered adequate in 6,276 bottle pairs. The total yield of isolates was 629, of which 489 (78%) were judged to be the cause of true infections. Staphylococci (P < 0.001) and yeasts (P < 0.05) were detected more often in PEDS PLUS bottles, as were all microorganisms combined (P < 0.001). The improved detection in the PEDS PLUS medium was most marked for patients on antimicrobial therapy (P < 0.001), bur remained statistically significant even for patients not on therapy (P < 0.025). There were 431 episodes of sepsis, including 407 considered adequate for analysis. Of the 363 unimicrobial episodes, 278 were detected by both bottles, 64 were detected by PEDS PLUS bottles only, and 21 were detected by AER bottles only (P < 0.01). No false-negative cultures were detected by terminal subculture of the PEDS PLUS bottles when the companion AER bottle was positive. However, there were 14 false-negative cultures (7 yeasts, 5 staphylococci, 1 Enterococcus faecalis, and 1 Enterobacter sp.) on terminal subculture of the AER bottles when the companion PEDS PLUS bottle was positive. When both systems were positive, the VITAL system detected bacteria earlier than did the BACTEC system by a mean of L6 h. Also, false- positive signals were less common with the VITAL system. We conclude that the VITAL system with AER medium must be modified to improve the detection of clinically important staphylococci and yeasts if it is to perform comparably to the BACTEC NR-660 nonradiometric system with PEDS PLUS medium for a pediatric population.