In this work, the application of statistical shape analysis to oropharyngeal structures from the population-based MRI data is investigated. For this purpose, statistical shape models (SSMs) of the relevant anatomical structures are created in order to determine the unknown parameters, which influence the shape of these areas. Subsequently, it is determined whether there is a connection between their shape and the occurrence of obstructive sleep apnea syndrome. Two statistical shape modeling approaches are investigated, namely, the classical SSMs constructed from the segmentation masks as well as (TL-)DeepSSM, which allows for extracting the shape models directly from the MRI scans without the segmentation process. The suitability of the methods for our particular application as well as their pros and cons are discussed. Additionally, the shape differences for healthy and diseased subjects using SSMs are presented.