Abstract: SNR (Signal-to-Noise Ratio) is an important parameter in communication signal processing and many algorithms need this information to obtain optimal performance. Firstly, signal power is calculated by determining spectral domain of Intermediate Frequency (IF) communication signal. Then, SNR is estimated according to its definition. Therefore, a fast blind SNR estimation algorithm is put forward for IF communication signals without any prior knowledge. Computer simulations are performed for the commonly used linear modulated communication signals, such as BPSK, QPSK, and 16QAM. Simulation results indicate that the algorithm is robust with estimated bias and root mean square error almost within 0.5dB when data length is 4096 and true SNR varies in the range from -5dB to 25dB. Compared with eigenvalue decomposition method, the proposed algorithm can achieve accurate estimation with low computational complexity and wide dynamic range.