A virtual-reality headset can put an astronaut outside a space station while they remain safely inside a training room. NASA uses simulations to rehearse situations that are difficult or dangerous to recreate on demand, including manoeuvres associated with spacewalks.¹ The attraction is clear: a person can practise decisions and movements in a convincing setting without needing the real spacecraft, operating theatre or rehabilitation environment every time.
That does not mean immersion is useful everywhere. A virtual scene is valuable when it lets someone do something they could not easily, safely or repeatedly do otherwise. The measure of success is what changes outside the headset: a skill learned, a decision improved or a movement recovered. Visual spectacle alone is a poor test.
Practice where mistakes are allowed
Training is a natural fit for VR because it can present repeatable situations. A learner can rehearse the same approach several times, receive feedback and try a rarer scenario that real life may never offer during a course. Astronaut training makes the logic vivid: a spacewalk is too consequential to treat the first real attempt as a rehearsal. NASA's training lab uses virtual simulations as one part of preparing crews.¹
Surgery offers another example, though the evidence needs careful reading. A systematic review of immersive VR for surgical training found promise as an adjunct for learning skills.² The word adjunct matters. A simulation can help teach hand-eye coordination, anatomy or procedural sequence, but it cannot reproduce every sensation, patient variation or team interaction in an operating theatre.
A useful test for a training programme has at least three stages. First, did the trainee improve inside the simulation? Second, does that improvement transfer to a real or realistic task? Third, does it endure and matter to the people affected by the work? Many demonstrations stop at the first stage. The later stages are more demanding and more relevant.
Rehabilitation is a different kind of rehearsal
After a stroke, rehabilitation may involve repeating movements that are frustrating or difficult to practise in an ordinary clinic. A virtual task can make repetition more engaging and can adjust its difficulty. But “VR therapy” covers a wide range of systems, some of which are interactive screens rather than fully immersive headsets.
A 2025 Cochrane review found that virtual reality and interactive video gaming may bring small improvements in arm function and balance compared with alternative therapy, and probably reduce some activity limitations. It also noted that relatively few included studies used head-mounted immersive VR, and confidence in the evidence varied.³ This is not proof that every headset-based product outperforms a skilled therapist.
One plausible benefit is simply more useful practice time. If an interactive task makes someone willing and able to perform additional repetitions, improvement may come from the extra dose of therapy as much as from being “inside” a virtual world. The Cochrane review found greater arm-function benefits when virtual methods were added to usual care, increasing therapy time.³ That distinction prevents technology from receiving credit for a treatment ingredient it merely helps deliver.
Design before anything is built
VR can also let designers and clients experience the scale of a proposed room or machine before it exists. A floor plan can specify dimensions precisely, yet a walk-through can reveal whether a doorway feels awkward or a workstation is hard to reach. Here the technology's advantage is spatial judgement, not a claim that the simulation is physically identical to the finished place.
The limits are just as important. A model can omit noise, texture, lighting conditions or the behaviour of other people. The headset may suggest confidence where the underlying design data are incomplete. A simulated building still needs engineering checks, accessibility assessment and consultation with those who will use it. Immersion can make a proposal easier to discuss; it cannot approve it.
This is a general rule for VR beyond entertainment. The virtual environment is a representation built from choices. Someone decides what to model, what to leave out and which outcomes to measure. A convincing view can help a learner grasp a problem, but it can also hide an omission behind visual polish.
A simulation can also expose disagreement productively. When two people view the same proposed space at human scale, they may notice different obstacles: a designer sees circulation, a wheelchair user sees reach, and a maintenance worker sees access to equipment. That conversation is valuable even if no one claims the digital model can replace a physical trial.
The body still has a vote
Headsets can cause discomfort. A 2025 systematic review of fully immersive VR in educational settings reported cybersickness and visual or balance-related symptoms in some studies, though the frequency and severity varied.⁴ Designers need to consider session length, movement, visual stability and whether users can stop easily. Access also matters for people who cannot comfortably wear a headset or who need a different interface.
Cost is another constraint. Hardware needs cleaning, maintenance and updates; specialist scenarios require careful software development. If a simpler screen, physical model or supervised practice achieves the same result, the headset may add expense without meaningful benefit. A fair comparison asks what would otherwise have been done with the same staff time and budget.
VR's most credible future is therefore less like a single new universe and more like a set of specialised tools. It can rehearse a dangerous task, support a programme of repeated movement or expose a spatial problem early. Each claim should be tested against real-world outcomes. The best use of a virtual world is to make someone better prepared for the actual one.