Abstract: The surrounding rock of deep engineering shows obvious creep characteristics. When it is affected by the external cyclic impact load with medium-grade energy (impacting energy of 10³~10⁵ J), a long-term large deformation dynamic disaster will occur. Based on the recent researches on the theory of creep disturbance effect of rock, the long-term large deformation mechanism and the stability control of deep dynamic engineering under cyclic medium-grade impact load are studied. The comprehensive research methods include the rock dynamics experiment, theoretical analysis of dynamic deformation and failure of the surrounding rock combined with the field engineering practice are adopted. According to the researches, the deformation law of creep disturbance of rock is summarized and analyzed. The stress and strain threshold indexes of creep rock mass sensitive to the external impact disturbance are determined. According to the relationship between the in-situ stress gradient and the strength gradient of surrounding rock, the state zone of the surrounding rock in deep dynamic engineering is re-divided. A new understanding about the mechanism of long-term large deformation and instability in deep surrounding rock is discovered based on the dynamic movement of e sensitive zone of disturbanc within the surrounding rock. The distribution and evolution law of the stress field in the urrounding rock is discussed. Then the long-term stability control principle of deep dynamic engineering is determined. That is, the supporting structures provide sufficient lateral confining pressure at the boundary of the sensitive zone of the disturbance. The lateral confining pressure must increase the anti-disturbance strength gradient in this zone to the static concentrated stress level. It can make the sensitive zone of disturbance disappear. The design method for support of the surrounding rock for deep dynamic engineering is also proposed based on the theoretical analysis, laboratory testing and field monitoring of dynamic load. The designing process of supporting parameters is optimized. The research findings have been applied in multiple field engineering examples with ideal results.