Described are various embodiments of an acoustic upper airway assessment system and method, and sleep apnea assessment system and method relying thereon. In one embodiment, a non-invasive method is described for assessing upper airway anatomy in a subject while breathing. This method comprises receiving as input at a hardware processor a digital signal representative of respiratory sounds generated by the subject while breathing； isolating digital respiratory sound segments from said digital signal； computationally extracting a designated spectral feature from each of said segments to characterize said respiratory sounds within each of said segments； automatically evaluating each said extracted spectral feature against a preset upper airway anatomy metric associated with said designated spectral feature to characterize a given upper airway anatomy measure in the subject while breathing； and outputting a sleep apnea severity indication based on said characterized upper airway anatomy measure.
