To compare the efficacy of oral gentamicin with nalidixic acid in the treatment of acute shigellosis, we studied, in a double blind-trial, 79 comparable children with bloody diarrhoea of less than 72 h duration. Of them Shigella spp. were isolated in 71 patients. Patients were randomly assigned to receive either gentamicin 30 mg/kg/day or nalidixic acid 60 mg/kg/day, both given orally in four equal doses for 5 days. Stool frequency differed significantly between the groups from day two until completion of the study. Treatment failure was observed in 14 (42 per cent) patients receiving oral gentamicin compared to none of those with nalidixic acid-sensitive strains of Shigella spp. (P = 0.0002). Although all the shigella isolates were sensitive to gentamicin in vitro, 19 (58 per cent) patients on gentamicin therapy failed to eliminate shigella organisms from stool, compared to none in the nalidixic acid treated group infected with nalidixic acid-sensitive Shigella spp. (P < 0.001). One patient in each group had a bacteriological relapse. We conclude that gentamicin given orally was therapeutically ineffective in the treatment of acute shigellosis.