Fifty children with bacterial meningitis were prospectively evaluated in a randomized comparative trial of twice daily ceftriaxone with conventional ampicillin and chloramphenicol therapy. The groups were comparable in age, sex, days of illness before admission, severity of illness at admission, etiology and admission cerebrospinal fluid (CSF) parameters and bacterial colony counts. The pathogens were Haemophilus influenzae type b (34 beta-lactamase-negative, 8 beta-lactamase-positive); Streptococcus pneumoniae (4); Neisseria meningitidis (3); and Streptococcus agalactiae (1). Initial CSF colony counts ranged from 2.5 X 10(2) to 1 X 10(10) colony-forming units/ml. In 44 children a lumbar puncture was repeated 10.5 to 18 hours after starting treatment; 16 of 24 (67%) ceftriaxone patients and 12 of 20 (60%) conventional therapy patients had sterile cultures. The reduction in the CSF bacterial colony counts (6.3 log10 colony-forming units/ml) was similar in both groups. Ceftriaxone CSF levels ranged from 1.0 to 8.0 micrograms/ml, representing a mean CSF penetration of 11.3% (range, 3.0 to 24.5%) of the simultaneous serum concentration. The median ceftriaxone bactericidal titer in CSF was 1:1024 compared with 1:4 achieved with conventional therapy. There were no significant differences in clinical responses or in frequency of complications, except for diarrhea which occurred in 59% of the ceftriaxone group and in 22% of the other (P less than 0.01). Despite one H. influenzae type b relapse occurring in the ceftriaxone group, ceftriaxone appears to be safe and as effective as conventional therapy for bacterial meningitis in children older than 2 months of age.