Abstract: The collapsibility of aeolian sand poses significant challenges to oil and gas resource development and the construction of related infrastructure in desert regions. This study focused on aeolian sand distributed along the southern margin of the Mu Us Desert. A combination of in-situ immersion load tests and laboratory physical property analyses was conducted to investigate the collapsibility characteristics and deformation behavior of the sand. The relationships between collapsibility and variables such as water content, dry density, porosity, and saturation were analyzed. Furthermore, a multiple linear regression (MLR) method was established to quantify the relationship between major physical indices and the collapsibility coefficient. It shows that the collapsibility of aeolian sand in this area is mainly slight, with a small proportion classified as non-collapsible. The load-settlement curve obtained from in-situ immersion loading test exhibits a polygonal pattern, which can be divided into three distinct stages: the linear elastic deformation stage (0−100 kPa), the elastoplastic deformation stage (100−200 kPa), and the collapse deformation stage (>200 kPa). Similarly, the immersion-induced settlement-time curve demonstrates a segmented characteristic, consisting of a rapid collapse phase during immersion (0−150 min) followed by a deformation stabilization phase. Notably, the collapse deformation shows a dramatic increase during the initial 150 minutes of immersion, then progressively stabilizes with prolonged immersion duration. The collapsibility coefficient of aeolian sand presents a negative correlation with initial water content, saturation, and dry density, but a positive correlation with porosity. The collapsibility evaluation model of aeolian sand based on multivariate linear regression can predict the collapsibility coefficient well, with a fitting degree of 0.879. In the future, the universality and accuracy of the model can be further improved by expanding the sample size and introducing nonlinear regression methods. These results provide a theoretical foundation for engineering projects in the Mu Us Desert and offer valuable insights into the collapsibility behavior of aeolian sand at similar sites.