<p>As an emerging video technology, 360-degree virtual reality (VR) video has garnered increasing attention due to its potential to create an immersive learning experience. However, the adoption of this immersive technology faces challenges related to the educational design and content production, and it remains unclear how the viewing environment of VR videos and learners’ prior knowledge influence learning. This study employs a two-stage research design to address these issues. In Stage 1, an embodied agent-based VR lecture system is developed through an iterative design process to facilitate the effortless creation of VR lecture videos, and a case of a geography VR lecture video was generated to demonstrate its benefits. The usability evaluation supports the potential of the system for creating usable VR lecture videos. In Stage 2, a total of 60 participants with varying levels of prior knowledge viewed the generated VR lecture video using a head-mounted display (HMD, <i>n</i> = 30) or a desktop screen (DS, <i>n</i> = 30) to explore the effects on learning performance and perceptual experiences. Results showed that (a) learners with high prior knowledge reported better satisfaction, presence, and agent evaluation than those with low prior knowledge; (b) learning in VR lecture video via HMD has also enhanced satisfaction, presence, and agent evaluation than DS, and (c) these positive effects were not moderated by the learner’s prior knowledge. The results indicate that the learning benefits of agent-based VR lecture videos brought by high-immersive viewing environments are applicable across learners with different prior knowledge levels. Overall, this study provides an educational design solution for VR lecture videos, and the findings underscore the value and general usability of this solution in assisting educators in developing immersive instructional media suitable for learners with diverse prior knowledge.</p>