Abstract The electrically evoked auditory brainstem response (eABR) is one of the clinically employed objective evaluation tools for cochlear implant (CI) subjects. It is commonly obtained by averaging responses, but because of the electric CI stimulation, some artifacts are phase locked to the stimulus and do not average out by increasing repetitions. A series of artifact reduction methods, such as general post-processing procedures for all subjects and individual post-processing procedures for some subjects, were developed in this study, aiming at reducing CI stimulation coherent artifacts. Seven bilateral CI subjects were recruited, and both monaural and binaural multi-channel eABRs were recorded in this study. The results show that the CI stimulation pulse artifacts can be efficiently removed by the general post-processing procedure, using alternating polarity stimuli combined with linear interpolation. Recordings obtained with non-alternating polarity show a strong exponential decay. Exponential fitting and subtraction worked reasonably well in this case. For eABR recordings contaminated with facial nerve stimulation (FNS) artifacts, principle component analysis was introduced to minimize the FNS artifacts for potential clinic application in the future.