R. Mayer has proposed the design principles of multimedia materials based on a cognitive theory of multimedia learning. In this study, we target to verify the modality principle and redundancy principle that Mayer mentioned, and consider individual differences of learners at the same time, to: 1) investigate the effect of multimedia presentation modes and learners' perceptual preferences on learning performance; 2) provide recommendations for design and selection of multimedia materials according to the results of this study. This study uses 3×2 two-factor quasi-experimental design. The participants are 83 undergraduate students. The independent variables in the experiment are the modes of multimedia presentation (animation+text, animation+narration, and animation+narration+text) and participants' perceptual preferences (visual, auditory). The material in the experiment is operated automatically, and the dependent variable is learning performance. In the first stage of this study, researchers collected data of pretest and perceptual preference questionnaire, and chose participants accordingly. Then participants received experimental material and posttest in the second stage. The main results of this study are: For the visual-perference-type and auditory-preference-type learners, 1) The multimedia presentation modes significantly affect learning performance, which is consistent with Mayer's arguments. The posttest scores from the highest to lowest are the animation+narration group, the animation+text group, the animation+narration+text group. The learners in both the animation+nairation group and the animation+text group spend less time than those who ate in the animation+narration+text group. 2) The perceptual preferences of learners have no significant effect on learning performance. 3) The multimedia presentation modes and learners' perceptual preferences have no significant interaction effect on learning performance.