Abstract: In order to establish a comprehensive and objective quality evaluation model, 13 kinds of salt coagulated tofu from large chain supermarkets, direct shopping stores and small retail stores were used as raw materials, and their sensory hardness, flavor, taste, color, sensory springiness, overall acceptability, texture hardness, texture springiness, cohesiveness, gumminess, chewiness, resilience and moisture state (water relaxation time and proportion) were measured. Single factor analysis of variance was used for multiple comparisons of each indexes, and bivariate test was used to analyze the correlation between sensory evaluation and instrument evaluation. Finally, a tofu quality evaluation model based on 18 quality indexes was established using principal component analysis (PCA). The results showed that there were significant differences in flavor, color, taste, overall acceptance, texture hardness, cohesiveness, gumminess, chewiness, resilience, relaxation time and peak ratio (except the ratio of bound water peak) among gypsum tofu samples. There were significant differences in quality indexes (except sensory springiness) among brine tofu samples. Tofu texture index was significantly correlated with sensory hardness, taste, color and overall acceptability, but was not significantly correlated with flavor and color. And T₂ relaxation time was significantly correlated with sensory hardness, taste, and color. Meanwhile, the peak ratio was positively correlated with sensory hardness and taste. The results showed that the quality of tofu cannot be evaluated by a certain index, but should be evaluated comprehensively by combining sensory index and instrument index. Through principal component analysis, the first three principal components were extracted, and the total variance contribution rate was 81.36%, which can reasonably reflect most qualities of tofu quality. PC1 integrated the information of sensory hardness, texture hardness, gumminess, and chewiness, and PC2 integrated the information of resilience and sensory springiness, and PC3 integrated the overall acceptability and sensory springiness. After further processing with principal component analysis, the evaluation model of tofu quality was established. The relaxation time of immobile and free water, sensory hardness, taste, texture springiness and resilience in the tofu gel system greatly contributed to the quality of tofu. Using the evaluation model, the tofu quality score of tofu can be calculated. The higher the score, the better the quality of tofu. Therefore, brine tofu has better sensory hardness, texture hardness, gumminess and chewiness, while gypsum tofu has higher scores in taste, relaxation time of immobile, and free water. This study provides a theoretical basis and method for the objective evaluation of salt-induced tofu sold in the market.