Abstract: Soil compaction risk can be normally assessed, in terms of the relationship between the wheel contacting pressure and the soil pre-consolidation stress. It is still lacking in the evaluation of the damage degree of soil structure that is caused by quantitative mechanical compaction in production scenarios, such as wet tillage and rot under intensive rice farming. In this study, a prediction model of soil compacted bulk density was derived using soil rebound and compression index in paddy fields under intensive production conditions. The improved model was verified by the laboratory uniaxial compression test and in situ flat subsidence test of undamaged soil in paddy fields. An indoor uniaxial compression test was also carried out under the undisturbed soil with different moisture content (15%, 20%, 25%, 30%, and 35%). Among them, there was no change in the bulk density of the undisturbed soil. The transfer functions were constructed for the soil initial bulk density (ρ), initial moisture content (w), elastic compression modulus (Es), plastic compression modulus (Ec), and soil pre-consolidation stress (σpc) after the uniaxial compression test. The coefficient of determination was greater than 0.95 after fitting the test data with each transfer function. The results show that the improved model was operable and accurate using the transfer function. At the same time, the Es presented a significant negative correlation with the ρ, whereas, there was a significant positive correlation with the w. The Es reached the maximum when ρ=1.1 g/cm3 and w=35%. The minimum was obtained, when ρ=1.8 g/cm3 and w=15%. The Es value ranged from 0-0.25 cm/kPa. The Ec presented a significant negative correlation with the ρ, and there was a quadratic polynomial relationship with the w. The Ec reached the maximum when ρ=1.1 g/cm3 and w=25%. The minimum Ec was obtained, when ρ=1.8 g/cm3 and w=15%, where the value ranged from 0-0.8 cm/kPa. There was a significant positive correlation between the σpc and ρ, whereas, a significant negative correlation was found between the σpc and w. The σpc reached the maximum when ρ=1.8 g/cm3 and w=15%. The minimum was obtained, when ρ=1.1 g/cm3 and w=35%. The value ranged from 20-160 kPa. The earth pressure was determined as 30, 60, 90, and 120 kPa in the in-situ flat subsidence test, according to the several types of harvesters. The soil bulk density in the uncompacted area was taken as the initial bulk density, while, the measured bulk density of the soil in the compacted area was as the measured bulk density. The ρ, w, elastic compression modulus, plastic compression modulus, and pre-consolidation pressure from the uncompacted region were then input into the prediction model of soil compacted bulk density to obtain the predicted bulk density. Finally, the measured bulk density was compared with the predicted. In-situ plate sinkage test showed that there was less than 5% error between the measured and the predicted using the transfer function-derived soil elastic compression and plastic compression modulus. At the same time, it was found that the wheel contacting pressure was greater than the soil pre-consolidation stress under the large w and the small ρ. There was a risk of soil compaction, even with the small contacting pressure of the wheel in the harvester. Therefore, a reasonable time and machine type can greatly contribute to the implementation of field tillage. Consequently, the prediction model of soil compaction can be expected to accurately quantify the soil bulk density under mechanical compaction. The finding can provide a strong reference for a better prediction model in regional agricultural usage.