Virtual reality (VR) has many uses across diverse areas such as scientific visualization, flight training, and architecture design. The spatial awareness and feeling of presence created by immersive virtual environments (IVE) assists with learning and reasoning tasks within VR and is one reason for its adoption. However, there are still problems of perception within IVEs that can limit the effectiveness of their use. One example is the compression of egocentric distances when using a head-mounted display to view the IVE. The goal for this dissertation is to investigate whether IVEs can be made more effective through the development of enhanced locomotion and interaction methods that provide more accurate visual and vestibular feedback to the user. We investigate the use of color and depth (RGB+D) cameras to generate real-time video-based self-avatars that perfectly match the user's own body without the need for markers or per-user calibration. We perform a user study to determine if the video avatars reduce the errors in egocentric distance perception experienced in virtual environments. Next, we discuss the geometric and perceptual errors present in multi-viewer single-view virtual reality displays, such as CAVEs. Since the calibration and setup of these displays is crucial to limit these errors, we have created open source software for calibration of these displays. Finally, we look at walking in IVE and discuss the benefits of natural movements over indirect interfaces such as keyboards and introduce the novel technique of redirected driving. We show that redirected driving has the potential to offer the benefits of walking, such as better spatial understanding and mapping recall, while still allowing movement of large distances.