Abstract: To mitigate mutual interference among waveforms during the cooperative detection process in multiple-access active sonar systems, a multiple-access waveform filtering technique based on the fractional Fourier transform (FrFT) is proposed. Firstly, according to the frequency coding characteristics of each subwave of the discrete frequency coding waveform of orthogonal frequency division linear frequency modulation (DFCW-OFD-LFM) signal, the energy focusing property of FrFT for subwaves with different coding modes, and the preset relative Doppler velocity range of targets, the rotation angles and FrFT-domain energy peak positions of each target subwave under different Doppler compression factors are calculated. On this basis, the waveform Doppler compression factor and the optimal FrFT rotation angle of each subwave are estimated. Subsequently, the FrFT-domain data and filtering window corresponding to the optimal rotation angle of each subwave are determined, and the interference power outside the filtering window of each subwave is eliminated to suppress waveform mutual interference. Finally, inverse FrFT is performed on the filtered subwave data, and the complete waveform data is reconstructed in accordance with the subwave sequence. To validate the performance of the proposed filtering method, simulation analyses are conducted to investigate the echo interference characteristics and multiple-access filtering characteristics under diverse mutual interference conditions. Furthermore, the replica correlation (RC) detection algorithm is adopted to simulate and verify the detection performance and miss detection performance of target waveforms before and after filtering at different Doppler velocities. The simulation results show that, at a false alarm rate of 1/1000 and a target detection probability of 0.8, the proposed method yields a detection performance gain of 0.91–3.07 dB at target relative Doppler velocities of 10 kn, 20 kn and 30 kn. In the presence of waveform mutual interference, the miss detection performance is improved by 2.07–5.10 dB at a miss detection rate of 0.1, which demonstrates the efficacy of the proposed filtering method.