Our primary objective was to assess diagnostic accuracy for acute otitis media (AOM) relative to the criterion standard established by the United States Agency for Healthcare Research and Quality: middle-ear effusion (MEE) plus onset in the past 48 h of signs or symptoms of middle-ear inflammation. A secondary objective was to assess the potential reduction in antibiotic usage that could be achieved if clinicians managed AOM according to a consensus guideline developed by the New York Region Otitis Project (NYROP). A convenience sample of primary care practitioners were surveyed after diagnosing AOM in 135 children aged 0.3-11.8 years (median 2.4 years). Clinicians expressed high certainty for AOM diagnosis in 122/135 episodes (90%). The prevalence of true AOM was 70% with a positive predictive value for high certainty of 76%. Of the 40 false-positive diagnoses, 35 did not have MEE and 5 did not have acute signs or symptoms. The relative risk for receiving an antibiotic was 1.50 times higher when clinicians expressed certainty (P=0.005), which produced 31/120 (26%) potentially unnecessary antibiotic prescriptions. Initial antibiotics would not have been prescribed for 29% of episodes using the NYROP guidelines. More judicious use of antibiotics may result if clinicians deferred initial therapy in children without definitive AOM, particularly when the presence of MEE is uncertain.