Abstract: This study aims to examine and compare the capability of different soundscape classification methods in explaining the physiological effects of soundscapes. Twenty representative soundscape samples of urban public open spaces are selected, and physiological measurements along with subjective evaluations are conducted on 62 participants. The research employs categorical principal component analysis to extract physiological dimensions and uses K-means clustering algorithm to classify both acoustic parameters and subjective evaluation data. A comparative analysis is performed to examine the explanatory power of three types of classification methods (by sound source, by acoustic parameter, and by subjective perception) on physiological dimensions. The results demonstrate that: 1) Physiological Dimension 1 (“comfort–discomfort” dimension) exhibits significant discriminability, effectively distinguishing soundscape differences across all classifications; 2) All three classification methods show limited differentiation effects on Physiological Dimension 2, with subjective evaluation classification demonstrating relatively better performance in distinguishing “vibrant-type” soundscapes. Overall, the discriminable information of soundscape physiological effects is mainly concentrated in comfort-related responses, whereas other physiological information shows generally weak sensitivity to classification methods.