Abstract: This study was aimed to explore the effects of tangential flow and axial flow mechanical threshing on number of mold colonies of Japonica rice,and the correlation between number of mold colonies of Japonica rice and seed quality.In this study,Japonica rice harvested by tangential flow and axial flow mechanical threshing patterns was used as mold colonies count test and germination experiment materials,Japonica rice harvested by manual was used as control group. Glume-opening rate in Japonica rice,number of mold colonies of Japonica rice,number of mold colonies of brown rice,and number of mold colonies of all glume-open and none glume-open Japonica rice under different threshing patterns were measured. The experiment results showed that there was significant difference in number of mold colonies of Japonica rice between the two different mechanical threshing patterns(P<0.05),number of rice mold colonies by tangential flow were more than 10⁵ CFU·g^-1,number of rice mold colonies o by axial flow were close to 10⁵ CFU·g^-1,while number of rice mold colonies by manual were slightly higher than 10⁴ CFU·g^-1. Number of rice mold colonies and number of brown rice mold colonies was significant linearly correlated with glume-opening rate(P<0.001),the determination coefficients between number of rice mold colonies and glume-opening rate,and between number of brown rice mold colonies and glume-opening rate were 0.9082 and 0.8719,respectively. The negative correlation coefficients between number of rice mold colonies and germination potential,and between number of rice mold colonies and germination rate were 0.917 and 0.905,respectively;while the positive correlation coefficients between number of rice mold colonies and the number of aberrations,and between number of rice mold colonies and the number of mildew in seedling root were 0.896 and 0.959,respectively. Therefore,in order to reduce number of rice mold colonies and maintain the seed quality,the axial flow mechanical threshing harvesting pattern should be preferred.