Objective: In order to study the correlation between fruit firmness, total soluble solids, red and green color, sensory scores of fresh yellow peaches, a prediction model of safe storage period of yellow peach was established to realize early warning and prediction of fruits quality. Methods: Solute Jinxiu peach was used as test material,firmness with skin, firmness without skin, fruit color difference,soluble solid contentand other indexes of the fruits in storage in three picking periods were measured, and sensory scores were made in four modules: fruit texture, color, flavor and aroma. The correlation analysis and time series analysis techniques were used to evaluate and establish a mathematical prediction model. Results: The best cluster number determined by contour curve is 4, which mainly showed the differences of fruit firmness, red-green color and soluble solids content, itwas consistent with the three module features of texture, color and flavor that consumers paid attention to.There is strong collinearity among sensory scores of the three modules of fruit firmness, color and flavor. Storage time is negatively correlated with fruit with skin and firmness without skin, and positively correlated with fruit red-green color difference.The red-green color of fruit is negatively correlated with fruit with skin and firmness without skin, which can be used as one of the characterization factors for nondestructive testing of firmness. Conclusion: At a temperature of (10.0±0.5) ℃ and a relative humidity of 80%~85%,a nonlinear prediction model of fruit with skin and firmness without skin was established: f[T.(a,k,b)]=a×exp(k×T)+b; The linear prediction model of red-green color of fruit was established: f(x)=kx+b. The verification results show that the prediction error of the above two prediction models is low (R²＞0.9, average error＜0.2).