Virtual Reality (VR) has progressed significantly since its conception, enabling previously impossible applications such as virtual prototyping, telepresence, and augmented reality However, text-entry remains a difficult problem for immersive environments (Bowman et al, 2001b, Mine et al , 1997). Wearing a head-mounted display (HMD) and datagloves affords a wealth of new interaction techniques. However, users no longer have access to traditional input devices such as a keyboard. Although VR allows for more natural interfaces, there is still a need for simple, yet effective, data-entry techniques. Examples include communicating in a collaborative environment, accessing system commands, or leaving an annotation for a designer m an architectural walkthrough (Bowman et al, 2001b). This thesis presents the design, implementation, and evaluation of a predictive text-entry technique for immersive environments which combines 5DT datagloves, a graphically represented keyboard, and a predictive spelling paradigm. It evaluates the fundamental factors affecting the use of such a technique. These include keyboard layout, prediction accuracy, gesture recognition, and interaction techniques. Finally, it details the results of user experiments, and provides a set of recommendations for the future use of such a technique in immersive environments.