Virtual Reality (VR), particularly head mounted display systems, impose different demands on people as compared to the real-world when performing a visuomotor task. One example is in visual perception, which has known impacts on visuomotor coordination. Although VR has been shown to have effects on how people perform visually guided skills, the technology is being used for visuomotor skills training, in both sport and non-sport contexts. There are several reasons for the use of VR technologies for skills practice and training, namely that skills can be performed, studied, and optimised in ways that are difficult to achieve in the real-world. The purpose of this thesis was to explore the use of VR for visuomotor skill evaluation, practice and training, and how the novel use of VR may improve real-world performance, independent of physical practice within a VR environment. Three studies are presented, each of which use the task of golf putting, a task selected based on a literature base with established performance evaluation metrics, related to both putting and visuomotor coordination as measured by eye-gaze. Study 1 aimed to evaluate whether real-world skill expertise when putting in VR environments is similar to putting in the real-world. Participants novice to golf (n = 46), split into low and high performing groups (LPN and HPN), national and international athletes in a sport other than golf (n = 14) and experts in golf (n = 5) putted in VR and in the real-world. The main measure of performance was radial error (the distance between the resting position of the ball and the hole). The results showed that performance is better in the real-world than in VR, while the real-world performance advantage of experts was reduced in VR. Quiet Eye, a gaze behaviour linked to improved performance, predicted a lower radial error when putting in the real-world, but not in VR. The results suggest that visually guided skills may be impaired in VR, and that VR testing environments may need to account for visuomotor differences to the real-world when evaluating performance or if VR is used for the practice of visuomotor skills (such as for repetitive golf putting). Study 2 examined Quiet Eye Training (QET), which aims to teach the QE fixation behaviour of experts. It evaluated how practicing golf putting in VR or in the real-world, with or without first receiving QET, affects skill performance pre-to-post and at a 1-week retention test. Golf novices were randomly divided into practice environment (real-world, RW; VR) and training type (QET, Control) groups to create four conditions: RW Control (n =16), RW QET (n = 11), VR Control (n = 10), and VR QET (n = 9). Receiving QET, irrespective of whether subsequent practice occurred in VR or in the real-world, improved real-world putting performance pre-to-post and at a longer-term retention interval, which was associated with more expert like gaze behaviours. Participants who practiced in the real-world, irrespective of training method, improved in their VR performance, while those who practiced in VR did not improve when tested in VR. The results suggest that VR practice alone may be unsuitable for skill learning and may need to be integrated within a training framework, such as QET, to produce real-world benefits to performance. Study 3 evaluated how showing a novice the eye-gaze behaviour of an expert golfer, without providing technical instructions or allowing for physical practice, affects gaze behaviour and putting performance in a pre-post design. The experts eye gaze was shown through an orange circle, displayed through a VR headset. One video showed the gaze cursor only (Cursor; n = 14), another with the peripheral information blurred (Blurred; n = 14), while the final condition was a passive Control group (n = 14). Cursor participants improved in radial error, showed a reduction in putt length, and had a better putt line pre-to-post. Blurred participants had a lower radial error, while Cursor participants did not show any performance improvement. QE duration increased for all conditions pre-to-post, with the largest increase for the Cursor group, followed by Blurred, then Control, while there was no difference between groups at post-test. The results show that observing the eye gaze of an expert through VR can assist visuomotor skill learning, independent of any further instruction or practice, which may expedite visuomotor skill acquisition. Over the three studies, the evidence suggests that perception and action may be poorly coupled in VR, at least when compared to the real-world. The inferior coupling may have flow on effects for how practice and training in VR should be conducted. The results suggest that an effective way to use VR is to show people how to perform skills in ways that are difficult to achieve in the real-world, as physical practice or further technical instruction beyond observation of an experts' eye-movements may not be required to produce benefits to real-world motor skill performance. When wanting to use VR to practice a visuomotor skill, it is suggested that a skill learning framework, such as QET, may be required in addition to practice to produce benefits to real-world skill learning. Training specific skills related to performance, such as QE, may benefit from being taught implicitly (such as through observation) and within the context of a wider visuomotor gaze sequence.