In the intelligent control system of magnetorheological (MR) suspension of all-terrain vehicle(ATV), the low-frequency disturbance(LFD) in the measured feedback signal (acceleration) makes the controller unable to calculate the theoretical control force accurately. Especially, when the frequency of the LFD signal is close to that of the actual acceleration signal, the LFD can’t be filtered by designing a traditional filter. Based on the above questions, this paper proposes an incremental proportion integration differentiation (IPID) strategy to address the issue of LFD in the measured feedback acceleration signal of the MR suspension system of ATV. First of all, the model of 1/4 vehicle suspension in consideration of LFD is established, the source of LFD is analyzed which is due to the transformation of Coriolis acceleration under the condition of vehicle body pitch and roll. Next, a semi-active IPID controller is designed by utilizing differential derivation of discrete PID and semi-active principle to eliminate the LFD component mixed in the acceleration signal by making a difference. The particle swarm optimization (PSO) algorithm is utilized to optimize the parameters of the controller, which is then verified through numerical simulation. Subsequently, a real vehicle control experiment is carried out based on a 4x4 ATV equipped with the MR suspension system and implemented by DSP controller with the designed IPID algorithm. The effectiveness of the prosed method is evaluated under the speed 10km/h and E road. And the designed method is compared with the traditional PID control algorithm through simulation and experimentation to demonstrate the superiority and rationality of "filtering out" LFD signal and improving control effectiveness.