Abstract: When cohesion needs to be considered in many cases,the factor of safety by the DDA using the strength reduction techniques is considerably underestimated compared with the traditional numerical methods such as the limit equilibrium methods.Some suggestions to improve the solutions are proved to be useless by numerical tests,such as changing the way of meshing blocks,improving the solution of block stress fields by alternative approximation techniques and raising the number of contact points on the boundaries.Through investigating the DDA contact model,it is found that the possible reason for the above phenomenon may be that the DDA deals with cohesion in a way quite different from the rigid plastic model adopted by the traditional limit equilibrium methods.The rigid plastic model regards cohesion as a constant,while the DDA takes the cohesion as zero once the slipping between the blocks boundaries occurs.But judging the occurrence of the slipping fully depends on the accuracy of normal and shear contact forces.If the contact forces are inaccurately evaluated,a falsely judged slipping state leads to the corresponding cohesion to be improperly eliminated.But the penalty method by the DDA for the contact analysis and the dynamic iteration computation are very hard to ensure the exact solution of contact forces.Based on this point,the rigid plastic model is used by the DDA to compensate for possible wrong loss of cohesion.Numerical tests prove the values of the factor of safety can be obviously raised,and they are even comparable with those by the traditional methods.This proves the suggestion of applying the rigid plastic model into the DDA is of significance while evaluating the slope stability factor of safety.