The Oxoid Signal (Oxoid U.S.A. Inc., Columbia, Maryland) system was compared with the nonradiometric BACTEC NR-660 (Johnston Laboratories, Towson, Maryland) system for detection of bacteria in 2714 blood cultures. The volume of blood collected into 20 ml blood-collection tubes containing sodium polyanetholsulfonate (SPS) (Becton Dickinson, Vacutainer Systems, Rutherford, New Jersey) ranged from 10 to 20 ml with an average of 15 ml. Subsequently, equal volumes of blood were inoculated into each system. A total of 250 organisms was isolated (9.6%), of which 149 (5.5%) were considered significant while 111 isolates from 98 cultures (3.6%) were contaminants. Of the significant isolates 32.9% were aerobic Gram-negative rods, 53.0% aerobic Gram-positive cocci, 5.4% anaerobes, 7.4% yeasts, and two isolates of Neisseria meningitidis. Ninety-five isolates were recovered in both systems, 29 by Bactec only and 25 by Signal only. Of the isolates recovered there were no significant differences in detection between the two systems with the exception of anaerobes (p less than 0.005). The median detection times for many of the most commonly isolated organisms--Enterobacteriaceae, streptococci, and Staphylococcus aureus--were very similar in both systems, ranging from 14 to 21 hours. With the remaining organisms recovered, the median times in hours for BAC-TEC and Signal, respectively, were 31 and 47 for Staphylococcus epidermidis, 48 and 60 for Bacteroides, 39 and 168 for yeast, and 16.5 and 168 for N. meningitidis. Oxoid Signal compares favorably with the BACTEC system. Its main advantages are: (1) it requires no instrumentation; (2) it is characterized by ease of detection; and (3) it uses a single-bottle system.(ABSTRACT TRUNCATED AT 250 WORDS)