One Salmonella and four Escherichia coli isolates from patients with bacterial meningitis who had responded slowly, relapsed, or failed to respond to monotherapy with moxalactam were examined. For purposes of comparison, an E. coli isolate from one patient who had responded promptly to therapy was also studied. On testing, moxalactam had higher MICs and MBCs (two to four times) than cefotaxime or ceftriaxone for all isolates; the rates of killing of the isolates were dependent on the antibiotic concentrations used. At comparable multiples of the MIC, these isolates were generally killed more slowly by moxalactam than by cefotaxime or ceftriaxone. In addition, a reduction of 3 in the logarithm of the number of CFU per milliliter could be attained at far lower concentrations with cefotaxime or ceftriaxone than with moxalactam. The degree of concentrationrelated killing of bacteria produced by the beta-lactams appeared to correlate with the clinical responses of the patients. Furthermore, real differences appeared to exist among the third-generation cephalosporins, which were not evident by the MIC and MBC points alone but were evident in the concentration-related killing curves. Determination of a reduction of 3 in the logarithm of the number of CFU per milliliter after a 6-h incubation is suggested as the criterion for the screening of antibiotics for the therapy of gram-negative bacillary meningitis.Soon after their introduction, the newer third-generation cephalosporins were generally accepted as potential antibacterial agents of choice for the therapy of gram-negative bacillary meningitis (1,4,11). A recent survey of infectious disease consultants in the United States showed that these antibiotics were indeed being used for the treatment of meningitis cases involving a wide range of bacterial pathogens (C. E. Cherubin, manuscript in preparation).In marked contrast to previously published series, scattered reports of therapeutic failures have begun to appear in the medical literature (2,5,8,12 Susceptibility testing. The CSF isolates were tested for susceptibility to antibiotics in three ways. First, the isolates were tested by the macro-broth dilution method with an inoculum of 5 x 105 CFU in 1 ml (14). A 0.1-ml portion was subcultured from each tube after an 18-h incubation at 35°C, and the criterion of 99.9% killing was used for the MBC. Second, the method described by Eng et al. (6) was used to ascertain the effect of a large inoculum on the susceptibility results. For this procedure, the organisms were tested by the same macro-broth dilution method but at an inoculum of 5 x 107 CFU/ml. The MIC at the higher inoculum was determined by the highest drug dilution which inhibited turbidity formation in the tube, compared with a control tube containing 1% Formalin and the same inoculum (6). A 100-pl sample from each tube was subcultured after an 18-h incubation at 35°C, and the same criterion of 99.9% killing was used to determine the MBC. Third, the organisms were retested in CSF and in Mueller-Hinton broth by ...