Objective: To reconcile conflicting published reports concerning the absolute and comparative clinical efficacy of antimicrobial drugs for acute otitis media in children. Study selection: Articles were identified by MEDLINE search, Current Contents, and references from review articles, textbook chapters, and retrieved reports. Randomized, controlled trials of therapeutic antimicrobial drugs used in the initial empiric therapy for simple acute otitis media were selected by independent, blinded observers, and scored on 11 measures of study validity. Thirty English and three foreign-language articles met all inclusion criteria. Data extraction: Data were abstracted for an end point of complete clinical resolution (primary control), exclusive of middle ear effusion, within 7 to 14 days after therapy started. Data synthesis: The spontaneous rate of primary control-without antibiotics or tympanocentesis-was 81% (95% confidence interval, 69% to 94%). Compared with placebo or no drug, antimicrobial therapy increased primary control by 13.7% (95% confidence interval, 8.2% to 19.2%). No significant differences were found in the comparative efficacy of various antimicrobial agents. Extending antimicrobial coverage to include β-lactamase-producing organisms did not significantly increase the rates of primary control or resolution of middle ear effusion. Pretreatment tympanocentesis was positively associated with individual group primary control rates, negatively associated with the ability to detect differences in clinical efficacy and unassociated with resolution of MEE. Conclusions: Antimicrobial drugs have a modest but significant impact on the primary control of acute otitis media. Treatment with β-lactamase-stable agents does not increase resolution of acute symptoms or middle ear effusion; initial therapy should be guided by considerations of safety, tolerability, and affordability, and not by the theoretical advantage of an extended antibacterial spectrum. (J Pediatr 1994;124:355-67).