To determine the effectiveness of antimicrobial treatment for otitis media with effusion ("secretory" otitis media) in children.We report the reexamination of a previously published study by Mandel et al that evaluated the efficacy of a 2-week course of antimicrobials (amoxicillin trihydrate) with and without a 4-week course of an oral decongestant-antihistamine combination in a double-blind, placebo-controlled, randomized trial involving 518 infants and children with otitis media with effusion.At 4 weeks, amoxicillin efficacy as determined by a tympanometric criterion (P = .121) or by a measure of improvement in hearing (P = .311) was insignificant. Only by otoscopic judgment, which is shown to contain a systematic bias as used in this clinical trial, could an argument be made for a marginal efficacy of amoxicillin at the 4-week end point. Logistic regression analyses of the combined effects of treatment and prognostic factors showed no significant differences between placebo- and antibiotic-treated groups for unilateral effusions and for bilateral effusions. When subjects with unilateral and bilateral effusions were combined, the estimated efficacy of antibiotic treatment was 12.3% by otoscopy (P = .014) and 4.8% by tympanometry (P = .171). We also demonstrate the sensitivity of outcome to diagnostic measures used and provide statistical evidence questioning the validity of otoscopic observations in this study. Six weeks after the termination of amoxicillin therapy, the recurrence of effusion was two to six times higher in the amoxicillin-treated children than in those treated with placebo (P = .001), and resolution of effusion was not significantly different among antibiotic and placebo groups (13.6% and 11.3%, respectively; P = .477).Amoxicillin with and without decongestant-antihistamine combination is not effective for the treatment of persistent asymptomatic middle-ear effusions in infants and children.