【Objective】Cadmium (Cd) is the major pollutant in farmland soils of China, especially in South China, where the rice produced in certain areas is very high in over limit rate. It is, therefore, essential to perform soil Cd risk assessment and explore for major pathways of the food chain getting exposed to Cd pollution. Based on the point-to-point data of Cd in the soil and brown rice of a certain tract of paddy field in Changsha, the Food Health Standard for Rice (GB 2762-2012) and the research target of protecting 95% of the bio-species in the ecosystem, a threshold value for soil Cd was worked out using the biological concentration factor (BCF) and SSD (species sensitivity distribution) equations, quantitative relationships of HC₅ (hazardous concentrations) with soil properties and their prediction models discussed and studies done to compare the threshold value with the criteria for Cd in the Standard for Soil Environment Quality of China.【Method】Some foreign research institutions and scholars usually use BCF as an important parameter in assessing effect and variety of the pollutant, and HC5 as the concentration of a substance in soil being toxic to 5% of bio-species in the ecosystem, in other words, protecting 95% of the bio-species in the ecosystem. HC₅ is the scientific basis for setting up soil environmental quality standards. In the present study, HC₅ for Cd in soil was derived using the species sensitivity distribution method and the concept of BCF based on point-to-point field data from paddy soils. Prediction models for Cd crop safety thresholds based on soil properties were also developed and a function of soil physicochemical properties established.【Result】Results show that soil pH is the main soil factor affecting soil Cd food safety thresholds, and followed by organic matter (OM) content and total Cd (TCd), explaining 62.2%, 19.4% and 18.3% of the variation of HC₅ in paddy soil, respectively. The two-factor (soil pH and OM) prediction models can predict Cd crop safety thresholds quite accurately with determination coefficients (R²) ranging from 0.802 to 0.817, while the three-factor (soil pH, OM and TCd) prediction models are more accurate than the two-factor ones. With rising soil pH or TCd, HC₅ increases in value, but the trend varies significantly. When the number of factors rises from two to three, the influence of TCd on HC₅ decreases in degree, while the effect of pH does not vary much. In the test range, OM and HC₅ show a certain negative relationship between the two. Soil Cd food safety thresholds vary with soil physical and chemical properties. The variation may lead to the the problem of insufficient protection and excessive protection at the same time, when the thresholds are used for controlling total Cd content by a single stage. In this study, a dynamic and quantitative food safety threshold model for Cd in soil was established.【Conclusion】The prediction models developed in the present study prove that quantitative relationships exist between soil properties and risk thresholds. And all the findings in this study are beneficial to developing and implementing soil specific, scientific and reasonable risk control of Cd in paddy soil and formulating regional soil environmental quality standards.