Abstract: Abstract: Mechanical damage of Citrus reticulate Blanco has an important influence on the postharvest characteristics. The objective of this study was to determine the effect of load conditions on mechanical damage of citrus and the protection performance of different material for citrus. Citrus without damage was set as a positive control group and the obviously damaged citrus was set as a negative control group which were stored for 7 or 14 days. Compression, dropping and vibration, 3 common loads in harvesting and the postharvest treatment, were conducted. Different constant load sizes (10, 30, 50 N horizontal, the compression part was the equatorial area of citrus; 50 N vertical, the compression part is the top of citrus) were tested in the compression experiment based on TMS-PRO texture analyzer using creep programming. Different heights (90, 105, 120 cm) were tested in the dropping experiment based on the dropping test platform. Different positions (1st, 2nd, 3rd, 4th floor), time (0.5, 1, 1.5 h) and accelerations (0.8, 1.15, 1.4 g) were tested in the vibration experiment based on 7YGD-40 self-propelled single-track orchard conveyor. Mass loss rate and relative permeability of cell membrane of citrus were measured after the 3 different experiments. According to the experimental data, using the technique for order preference by similarity to ideal solution (TOPSIS), the data of positive control group were considered as ideal point, the degree of mechanical damage of citrus was defined and calculated, and the mechanical damage of citrus was quantitatively estimated based on the mass loss rate and relative permeability of citrus. Based on the k-nearest neighbor classification in machine learning, using scikit-learn in Python, the positive and negative control groups were used as training data set, the dichotomous prediction model of mechanical damage of citrus was established, and the damage rate of citrus was qualitatively classified. The test accuracy was over 80%. Due to the mass loss and the change of the cell membrane permeability under different load conditions, the 5 factors of the 3 experiments had extremely significant effects on the mass loss rate and relative permeability of citrus (P<0.01). The predictions of the mechanical damage of citrus showed that the pressure should be less than 10 N to keep the unprotected citrus from damage. It was suggested that the contact force should not exceed this value if the rigid manipulator was used to pick up the citrus. When vibrated, the number of stacked layers of the citrus should be decreased as much as possible, and it was suggested to be 2 layers if citrus was contacted directly, and the citrus on the upper layers should be provided a cushion to protect against shock. Vibration acceleration should be less than 0.8 g and vibration time should be less than 0.5 h. In addition, the evaluation of the damage degree of citrus showed: Corrugated paper had the best effect on reducing damage (damage degree was only 0.4%), and plywood had the secondary effect (damage degree was 4.77%) in the compression experiment; EPS (expanded polystyrene) had the best effect on reducing damage (damage degree was 6.46%) in the dropping experiment; HDPE (high density polyethylene) had the best effect on reducing damage (damage degree was 4.33%) in the vibration experiment. The results of the study can provide a reference for equipment design and damage protection in the harvesting and postharvest treatment of Citrus reticulate Blanco.