Background: There is currently no method that can predict whether or under what condition hypopnea, even obstructive sleep apnea (OSA), will occur during sleep for individuals based on credible parameters measured under waking condition. We propose a threshold concept based on the narrowest cross-sectional area of the upper airway (CSA-UA) and aim to prove our hypothesis on the threshold of the area for hypopnea onset (TAHO), which can be used as an indicator of hypopnea onset during sleep and measured while awake. Methods: We performed magnetic resonance imaging for 20 OSA patients to observe CSA-UA changes during fluid accumulation in the neck caused by elevating their legs, and identified TAHO by capturing the sudden enlargement in CSA-UA. Correlation analyses between TAHO and the body mass index (BMI), and between the reduction in CSA-UA and the increase in the neck circumference (NC) with fluid accumulation were performed. Logistic regression analysis was performed for identifying OSA patients based on the behaviors of their CSA-UA changes during leg raising. Shape changes of airway cross-section were also investigated. Results: Four CSA-UA change patterns after fluid redistribution were identified. Six patients had similar CSA-UA variation behaviors observed in healthy subjects. From the other three change patterns involving 14 patients, a threshold value of CSA-UA 0.63 +/- 0.21 cm(2) was identified for normal breathing. Data showed a positive correlation between TAHO and BMI (r = 0.681, p = 0.0007), and a negative correlation between the reduction in CSA-UA and the increase in NC (r = -0.513, p = 0.051) with fluid accumulation. A sigmoid function for the probability of being a OSA patient p = 1/[1 + exp. (4.836 + 3.850 t-8.4 h)] was obtained to effectively separate OSA patients from normal subjects. The upper airway narrowing occurred in anteroposterior, lateral, or both directions, suggesting different tendencies of upper airway collapse in patients. Three types of shape changes in the cross-section of the upper airway, which had different effects on airway resistance, were measured. Conclusions: Our findings prove TAHO hypothesis. The threshold measured while awake for normal breathing can be used clinically as the indicator of hypopnea onset during sleep, and therefore to identify OSA patients under waking condition and design effective personalized treatments for OSA patients. Both shape and size changes in the cross-section of the upper airway affect airway resistance significantly. Shape change in the cross-section of the upper airway can provide key clinical information on the collapse patterns of the upper airway for individuals.