One hundred five infants and children with acute otitis media were randomized to therapy with loracarbef, an experimental carbacephem antibiotic, or amoxicillin-clavulanate (Augmentin), an approved drug for this disease. Ninety-two were evaluable (46 in each group). Middle ear fluid samples obtained for culture before therapy grew Haemophilus spp. in 30% of cases, pneumococci in 29% of cases, and Moraxella catarrhalis in 15% of cases. beta-Lactamase-producing bacteria were found in 37% of patients. Clinical failure occurred in four loracarbef-treated patients and one amoxicillin-clavulanate-treated patient (P = 0.361). Recurrence of acute otitis media was more common in the 2 to 3 weeks after loracarbef treatment (eight patients) than it was after amoxicillin-clavulanate therapy (three patients), but not significantly so (P = 0.197). Thus, combined failure and recurrence occurred in 12 loracarbef-treated patients and four amoxicillin-clavulanate-treated patients (P = 0.052). Gastrointestinal side effects occurred in 13 loracarbef-treated and 21 amoxicillin-clavulanate-treated patients (P = 0.13). Diaper rash was more common with amoxicillin-clavulanate (22 patients) than with loracarbef (10 patients; P = 0.016). Satisfactory results were achieved with both antibiotics, and adverse effects, although common, were minor.