Background: The longer upper airway is more collapsible during sleep. This study aims to reveal relationships among upper airway length, weight, and obstructive sleep apnea (OSA), particularly to answer why the upper airway of OSA patients is longer than that of healthy people and why some obese people suffer from OSA while others do not. Methods: We perform head and neck MRI on male patients and controls, and measure >20 morphological parameters, including several never before investigated, to quantify the effect of weight change on upper airway length. Results: The upper airway length is longer in patients and correlates strongly to body weight. Weight increase leads to significant fat infiltration in the tongue, causing the hyoid to move downward and lengthen the airway in patients. The apnea-hypopnea index (AHI) strongly correlates to airway length and tongue size. Surprisingly, a distance parameter h and angle beta near the occipital bone both show significant differences between healthy males and patients due to their different head backward tilt angle, and strongly correlates with AHI. The contributions of downward hyoid movement and head tilt on airway lengthening are 67.4-80.5% and19.5-32.6%, respectively, in patients. The parapharyngeal fat pad also correlates strongly with AHI. Conclusions: The findings in this study reveal that the amount of body weight and distribution of deposited fat both affect airway length, and therefore OSA. Fat distribution plays a larger impact than the amount of weight, and is a better predictor of who among obese people are more prone to OSA.