The current gold-standard method of diagnosing obstructive sleep apnea (OSA) is polysomnography, which can be inefficient. We therefore sought to determine a method to triage patients at risk of OSA, without using subjective data, which are prone to mis-reporting. We hypothesized that acoustic pharyngometry in combination with age, gender, and neck circumference would predict the presence of moderate-to-severe OSA.Untreated subjects with suspected OSA were recruited from a local sleep clinic and underwent polysomnography. We also included a control group to verify differences. While seated in an upright position and breathing through the mouth, an acoustic pharyngometer was used to measure the minimal cross-sectional area (MCA) of the upper airway at end-exhalation.Sixty subjects were recruited (35 males, mean age 42 years, range 21-81 years; apnea-hypopnea index (AHI) 33 ± 30 events/h (mean ± standard deviation), Epworth Sleepiness Scale score 11 ± 6, body mass index 34 ± 8 kg/m(2)). In univariate logistic regression, MCA was a significant predictor of mild-no OSA (AHI < 15). A multivariate logistic regression model including MCA, age, gender, and neck circumference significantly predicted AHI < 15, explaining approximately one-third of the total variance (χ(2)(4) = 37, p < 0.01), with only MCA being a significant independent predictor (adjusted odds ratio 54, standard error 130; p < 0.01).These data suggest that independent of age, gender, and neck size, objective anatomical assessment can significantly differentiate those with mild versus moderate-to-severe OSA in a clinical setting, and may have utility as a component in stratifying risk of OSA.