Abstract: Secure Cognitive Radio (SCR) system based on Reconfigurable Intelligent Surface (RIS) assumes that Secondary User (SU) and Primary User (PU) locate on the same side of the Base Station (BS), which can only cover part of the communication area and limits the deployment flexibility and effectiveness of the RIS. In order to solve the above problem, a new SCR system based on Simultaneously Transmitting And Reflecting Intelligent Surface (STARS) was proposed in this paper. In the system, STARS can achieve full coverage of communication area, improved the received signal strength of the SU with the transmission beamforming vector and reduced the interference of the SU to the PU with the reflection beamforming vector, which provided new optimization Degree of Freedoms (DoFs) for the design of the SCR system. Under conditions of the secrecy rate constraint of the SU, Interference Power Constraint (IPC) of the PU and transmission/reflection parameters constraint of the STARS, BS active beamforming vector and STARS transmission/reflection beamforming vectors were jointly optimized to minimize the BS’s transmit power from the perspective of reducing the total power consumption of the system. The minimization problem was a variable coupling non-convex problem, which was difficult to tackle directly. Then, an Alternating Optimization (AO) algorithm based on Difference-of-Convex Relaxation (DCR) method and Sequential Rank-One Constraint Relaxation (SROCR) method was proposed to jointly design the BS active beamforming vector and STARS transmission/reflection beamforming vectors. Simulation results show that the proposed algorithm has good convergence performance and effectively reduces the interference of SU on PU. Compared with the traditional RIS, random phase, Maximum-Ratio Transmission (MRT) and Equal Energy Splitting (Equal ES) schemes, the BS transmitting power is reduced by 8.3%, 15.4%, 5.9% and 5.3%, respectively.