Ampicillin-cefotaxime was tested as initial therapy of presumptive bacterial meningitis in 55 children greater than or equal to 2 months of age at our hospital. During the first year of this ongoing trial, 11 patients, 10 whose CSF yielded ampicillin-resistant Haemophilus influenzae type b (MIC greater than 16 mg/l, beta-lactamase +) and one, indole-negative proteus (MIC 4 mg/l), were begun on ampicillin-cefotaxime and then continued on cefotaxime alone. All did well clinically except one who convulsed briefly but recovered without sequelae. The cefotaxime MICs/MBCs of the beta-lactamase-positive H. influenzae isolates (less than or equal to 0.007 to 0.03/less than or equal to 0.007 to 0.12) and the proteus isolate (0.03/0.12) were significantly lower than chloramphenicol MICs/MBCs (0.25 to 1.0/0.5 to 1.0 and 8/greater than 16). We followed 44 other children with meningitis due to ampicillin-sensitive organisms who were treated with ampicillin or penicillin after 1 or 2 days of ampicillin-cefotaxime. Aetiological agents included ampicillin-sensitive H. influenzae (25), pneumococci (9), meningococci (8), Strept. MG (1) and Listeria monocytogenes (1). 40/44 recovered uneventfully. There were 4 neurological complications: the streptococcal meningitis sustained a brain abscess and the three others were motor incoordination (sensitive haemophilus), hearing loss and subdural effusion (2 pneumococci). There were no deaths. 18/48 children managed initially with ampicillin-chloramphenicol during the same 12-month period one year earlier had significant neurological complications and/or sequelae and there was one death; aetiological agents included sensitive H. influenzae (30), pneumococci (9), ampicillin-resistant haemophilus (5), meningococci (3) and pneumococci plus strept. MG (1). The two groups were comparable except for the number of resistant haemophilus and meningococcal strains and underlying disease more frequent in the ampicillin-cefotaxime group. A significant reduction of neurological morbidity (5/55 or 9.1% vs. 18/48 or 37.5%:P less than 0.001) was therefore associated with the ampicillin-cefotaxime schedule in the initial treatment of proven bacterial meningitis. A prolonged hospitalization (greater than 15 days) was less frequent (P less than 0.01) in the ampicillin-cefotaxime group (3/55 or 5.5% vs. 13/48 or 27.1%). The results of the trial to date are considered to be very promising.