A single-blind therapeutic trial, using randomly either cefotaxime or a benzyl-penicillin-gentamicin combination, was carried out in 68 hospitalised paediatric patients with 72 episodes of severe infection, which were, in the main, septicaemia, pneumonia, neonatal meningitis and a few other miscellaneous infections. The cefotaxime group showed a cure rate of 94.4% compared with 72.2% in the other group. One patient with bacterial meningitis treated initially with cefotaxime died a month later; however, penicillin and chloramphenicol had been added due to clinical deterioration. In the penicillin-gentamicin group there were five deaths, all from suspected neonatal septicaemia, and three cases required a change in antibiotic regimen before a cure could be effected. The results indicate that cefotaxime should be considered a drug of choice in many neonates with life-threatening sepsis.