A total of 150 children with acute otitis media were randomly allocated to treatment with amoxicillin-potassium clavulanate (Augmentin) or with cefaclor. Each drug was given in a daily dosage of approximately 40 mg/kg in three divided doses for ten days. Tympanocentesis done before treatment yielded specimens that contained pneumococcus or Haemophilus sp or both in 67% of specimens. Viridans group streptococci were isolated from 10% of specimens and Branhamella catarrhalis from 6%. Patients were scheduled for follow-up examinations at midtreatment, end of therapy, and at 30, 60, and 90 days. Of the 150 children, 130 were evaluable. Five of 60 patients (8%) treated with cefaclor were considered therapeutic failures because of persistent purulent drainage and isolation of the original pathogen or suprainfection. There were no failures among patients treated with Augmentin (P = .019). Rates of relapse, recurrent acute otitis media with effusion, and persistent middle ear effusion were comparable in the two groups of patients. Diaper rash, or loose stools, or both were significantly more common in children treated with Augmentin (34%) than in those taking cefaclor (12%), but in no case was it necessary to discontinue medication because of these mild side effects (P = .002). Cefaclor therapy was discontinued in one patient because of severe abdominal pain and vomiting. In this study, treatment with Augmentin was superior to treatment with cefaclor in the acute phase of acute otitis media with effusion, but Augmentin produced more adverse effects. The rates of persistent middle ear effusion and recurrent acute otitis media with effusion were comparable with the two regimens.