Twenty-five published trials of antibacterial therapy for acute otitis media were reviewed according to 13 methodologic standards for the design of clinical trials. Randomized, controlled trials, often with the double-blind technique, have been widely applied to assess efficacy. Bias can be further avoided by prognostic stratification according to known risk factors and by the measurement and analysis of patient compliance with treatment. A less frequently recognized problem is the insensitive trial. Bacteriologic diagnosis of the middle ear exudate before therapy can improve the sensitivity of clinical trials. The eradication of pathogens from the middle ear may be a more sensitive measure of outcome than the resolution of ear effusion. The analysis of type II statistical error should be included in trials that reveal no difference between antibacterial agents. The evaluation of new antibacterial agents for this important health problem requires rigorous research design.