It is usually required that fingerprint images in the database of an AFIS(automated fingerprint identification system) should be compressed to improve its practical performance and also that the restored images should have high fidelity to avoid loss of minutiea while the time consumed in the restoration is required short enough. This paper proposes a high-performance compression method for the latent fingerprint images based on the FSWT(fast symmetric wavelet transform) which meet the requirement of high speed and high fidelity by simultaneously using the FSWT, the proper decomposition style, the adaptive scalar quantization and the improved entropy coding method. Meanwhile a bit rate control strategy is designed to control the compression ratio effectively. The experimental results show that the proposed compression method has high performance.