To improve global control of disease and reduce global toxicity, a complex seed distribution pattern should be achieved with great accuracy during brachytherapy. However, the interaction between the needle and prostate will cause large deformation of soft tissue. As a result, seeds will be misplaced, sharp demarcation between irradiated volume and healthy structures is unavailable and this will cause side effects such as impotence and urinary incontinence.In this paper, a 3D nonlinear dynamic finite element simulation method with application to robot-assisted image guided brachytherapy is proposed. A magnetic resonance imaging (MRI) based simulation model is built which considers real patient data, nonlinear material property and interaction between the needle and soft tissue. The sensitivities of seed placement error to the needle insertion point, needle orientation and insertion distance are analysed. The result shows that shorter distance between insertion point and target position with compensation in distance can reduce placement error effectively. This method can be used as a complementary instruction for robot-assisted image guided brachytherapy.