To compare the clinical outcome of treatment of cholera in children with ampicillin, erythromycin or tetracycline, a double-'blind' randomized four-cell trial was carried out in Bangladesh. Ampicillin was chosen as additional therapy for acute respiratory tract infection, present in many subjects with diarrhoea. One hundred and eighty-four children aged 1-5 years who were not wasted, with diarrhoea of duration < 48 h, signs of some or severe dehydration, dark-field stool microscopy demonstrating Vibrio cholerae, and a baseline purging rate > 4 mL/kg/h over 6 h were enrolled in the study. Ampicillin, tetracycline, erythromycin or placebo were given orally every 6 h for 3 d. After 3 d of antibiotic treatment, diarrhoeal stool volume was significantly reduced in all antibiotic groups, with mean volumes per kg body weight as follows: tetracycline, 318 mL (SEM = 50), ampicillin, 335 mL (SEM = 30); erythromycin, 323 mL (SEM = 25); placebo, 498 mL (SEM = 37). Compared to tetracycline, the clinical recovery rates by 96 h were 75% with placebo, 91.3% with ampicillin, and 95.7% with eythromycin. Compared to tetracycline, the total mean times to recovery were increased by 66% with placebo (P < 0.001), 25% with ampicillin (P < 0.017), and 9% with erythromycin (P = 0.37). These results indicated comparable clinical efficacy of tetracycline, ampicillin and erythromycin. We therefore recommend that, unless V. cholerae is resistant, ampicillin should be used as a cost-effective alternative to erythromycin for paediatric cholera, especially in children with concomitant acute respiratory infection.