Auditory brainstem response (ABR) is a clinical auditory detection tool that can evaluate the function of the central auditory pathways through the brainstem, but is easy to be interfered with by noise, which requires subjects to keep quiet during tests. However, it is hard for children or adults that cannot cooperate to keep quiet for such a long time. Besides, the ABR test is time-consuming because thousands of trials are needed. In this study, an adaptive Kalman filtering (AKF) method was proposed to help with the ABR acquisition in the motion (chewing or mouth open). We first studied the feasibility of the AKF method by manually adding noise to electroencephalogram (EEG) trials that were used to acquire ABR on adult subjects. Then, we compared the performance of AKF with the traditionally used averaging (Ave) and artifact rejection (AR). The results showed that the AKF-based ABR achieved 96.16 ± 2.15% of the correlation coefficient and similar morphology as the Ave-based method in rest. In motion, the AKF-based ABRs had more recognizable characteristic waves, stable latencies, and higher wave V's amplitudes than those of Ave or AR-based methods. It is believed that the AKF-based method provides the possibility of in-motion ABR acquisition.