Recent research has shown that cognitive radio (CR) is an attractive technology to enhance the spectrum utilization and the fractional Fourier transform (FRFT) is widely used for its advantages in dealing with the time-varying degradations and non-stationary processes. In this paper, a novel interference suppression approach using the CR-FRFT is proposed for the spread spectrum systems. This approach firstly carries out the pattern recognition by spectrum sensing, and then chooses an optimal transform order for the recognized interference to make the FRFT-based transform domain excision. Theoretical analysis and experimental results show that the proposed CR-FRFT technique is more pertinent than the conventional techniques and can achieve a significant performance for the broadband interference. As the FRFT has a fast digital algorithm, it is conveniently to make a software package of the proposed CR-FRFT technique and be embedded in the existed spread spectrum systems for interference excision applications.