In this paper, we study the preventive maintenance policy for a two-component series system. When the working time of the system reaches T, the system will be preventive repaired. The preventive maintenance restores every component to the state which is the same as the last failure repair. When the failure occurs, the repair may be delayed for every component. The successive survival times of the component form a stochastically decreasing geometric process and the consecutive repair times after component failures form a stochastically increasing geometric process. We look for a bivariate policy (T,N), which T is the working time of the system before preventive maintenance, N is the failure time of the component before replacement. By using the renewal process theory and geometric process theory, the explicit expression of the long-run expected cost per unit time is derived. Finally, a numerical example is given.