The influence of a prefermentative freezing process on changes of aromatic characteristics and volatile compounds in Meili wines was studied to optimize freezing parameters and reveal the mathematical relationship between aromatic characteristics and volatile compounds. The wines obtained were characterized by sensory evaluation and stir bar sorptive extraction (SBSE) followed by a thermal desorption-gas chromatography-mass spectrometry analysis. A total of 28 aromatic descriptors from 6 categories of wine aroma terminology were identified by judging with high "modified frequency (MF%)". In addition, 19 varietal aroma compounds and 36 fermentation aroma compounds were quantitated, followed by the determination of odor activity values (OAVs). On the basis of the data obtained, principal component analysis (PCA) was used to find the relationship between characteristic aroma terms and different freezing conditions, and then partial least-squares regression (PLSR) was proposed to establish the mathematical relationship between the resulting terms and impact odorants. Natural thawing treatment on frozen must resulted in higher aroma quality with higher extraction of varietal aroma compounds. Lower frozen maceration temperature contributed to higher esters and organic acids. Impact aroma compounds were related to models for floral, sweet fruit, temperate fruit, and vegetal, whereas the model of rose and strawberry contained only varietal volatile compounds, and temperate fruit could be regressed by impact fermentation aroma compounds.