In a multicentre study, 220 consecutive cases of bacterial meningitis in children older than 3 months were randomised to treatment with chloramphenicol, ampicillin (initially with chloramphenicol), cefotaxime, or ceftriaxone. The drugs were given in four equal daily doses for 7 days, except ceftriaxone which was given only once daily. 200 cases could be assessed; the causative organisms were Haemophilus influenzae type b (Hib) in 146; meningococci (Mnc) in 32; pneumococci (Pnc) in 13; and other or unknown in 9. In patients with Hib meningitis, sterilisation of the cerebrospinal fluid occurred most rapidly with ceftriaxone. Otherwise, in terms of overall clinical recovery, normalisation of laboratory indices, clinically significant adverse reactions, toxic effects, sequelae, and mortality rate, the treatment groups were very similar. However, there were 4 bacteriological failures, all in the chloramphenicol group. Also, the treatment was extended or changed in more cases in the chloramphenicol group than in the other groups. Chloramphenicol was thus inferior to the other three antimicrobials. Ampicillin is a good and cheap alternative, but there are difficulties with resistance. Easy administration tempts the use of ceftriaxone rather than cefotaxime but it causes diarrhoea. A 7-day course of ampicillin, cefotaxime, or ceftriaxone is sufficient in Hib, Mnc, or Pnc meningitis.