Abstract: An inverse Doppler shift algorithm based on ZoomFFT to measure Doppler mismatch of underwater communication signals is presented. Doppler mismatch is corrected and bit error rate is reduced with this method. Simulations on both synthetic and real signals show that this algorithm is approximately two orders of magnitude faster, as well as four orders of magnitude less memory storage than the traditional sliding correlation algorithm, and settles the Doppler shift compensation during the course of the mobile communications of underwater signals. The result shows the algorithm can be used in practical underwater communication.