Objective: We identified the pathogens causing persistent and recurrent acute otitis media (AOM) and the clinical efficacy of cefprozil as treatment. Study design: This was a noncomparative, open label multicenter trial. Children ages 6 months to 12 years with signs and symptoms of AOM and evidence of middle ear effusion, as confirmed by pneumatic otoscopy or tympanometry, underwent tympanocentesis and subsequent treatment with cefprozil (15 mg/kg given twice daily) for 10 days. Patients with recurrent otitis media or failure of previous antibiotic therapy or prophylaxis were particularly sought for the study. Results: Two hundred sixty-two (99%) of 265 enrolled children were considered evaluable. The median age of the study group was 1 year. Ninety-eight (37%) of the children had a history (within 30 days) of prior antibiotic use. Ninety-seven (37%) met our definition of recurrent AOM, 48 (18%) met our definition of persistent AOM and 132 (50%) children had 3 or more previous episodes of acute otitis media within 12 months before study. Eighty-two (31%) of the enrollment tympanocentesis had no growth, 150 (57%) had a single bacterial pathogen and 29 (11%) had multiple bacterial pathogens. Of the 93 Streptococcus pneumoniae pretreatment isolates, 50 (54%) were penicillin-susceptible, 12 (13%) were penicillin- intermediate resistant and 31 (33%) were penicillin-resistant. Of the 75 Haemophilus influenzae pretreatment isolates, 42 (56%) produced betalactamase as did 4 (27%) of the 15 Moraxella catarrhalis strains. A satisfactory clinical response by pathogen was found in 75% (70 of 93) with S. pneumoniae, 75% (56 of 75) with H. influenzae and 93% (13 of 14) with M. catarrhalis; the response with single pathogen infections was higher than those with multiple pathogens (118 of 150 (78%) and 17 of 29 (59%), respectively; P = 0.03). The response for patients with isolates of S. pneumoniae that were penicillin- susceptible, -intermediate or -resistant were 39 of 50 (78%), 11 of 12 (92%) and 21 of 31 (68%), respectively. Older children had a satisfactory clinical outcome more frequently than younger children (P < 0.001), and the response to therapy varied for persistent, recurrent and recently untreated AOM (P < 0.01). Conclusion: Persistent and recurrent AOM involves the same pathogens as recently untreated AOM but bacteria with reduced antibiotic susceptibility may be more frequently present. This noncomparative study suggests that cefprozil 30 mg/kg/day given in two divided doses for 10 days may be effective in the treatment of children with persistent and recurrent AOM.