Skip to content

SendorVR goes live Nov 2026.

Join the Waitlist

Science

Does Virtual Reality Actually Work for Phobias

Shabnam Tahernia

Written by

Shabnam TaherniaShabnam Tahernia on LinkedInHead of Content

Does Virtual Reality Actually Work for Phobias

Short answer: for specific phobias, yes — and it has held up better than most people expect. Across dozens of randomized controlled trials, virtual reality exposure therapy has produced large improvements compared with waiting lists, and has come out roughly level with the gold standard of facing the real thing in person.

The longer answer is where it gets interesting. "VR works for phobias" is a sentence doing a lot of quiet work. It hides which fears were studied, how the headset was used, who was in the room, and how long anyone was followed afterward. Those details are the difference between a finding and a headline.

What researchers mean by VR exposure therapy

Exposure therapy is an old, well-mapped treatment. A person with a phobia approaches the thing they fear in small, deliberate steps, stays with the discomfort long enough for it to settle, and repeats that until the fear loses its grip. It is uncomfortable by design.

Virtual reality exposure therapy, usually shortened to VRET, swaps the real spider or the real balcony for a simulated one. The appeal is practical rather than magical: a therapist can conjure a thunderstorm at 30,000 feet on a Tuesday afternoon, restart it, dial it down, and run it again — none of which is possible when the exposure requires an actual airport.

This matters for reading the research. VRET is a course of treatment aimed at lasting change, typically delivered or supervised by a clinician. It is a different thing from putting a headset on someone for three minutes while a procedure happens, which we have written about separately. Studies of one rarely tell you much about the other.

How strong is the evidence?

The most cited synthesis is a 2019 meta-analysis by Carl and colleagues, which pooled 30 randomized controlled trials covering 1,057 participants across specific phobias, social anxiety, PTSD and panic disorder. Compared with waitlist controls, VRET showed a large effect (g = 0.90). Compared with psychological placebo conditions, the effect was medium to large (g = 0.78). Compared with in-vivo exposure — the real-world version — the difference was essentially nil (g = −0.07). The authors concluded that VRET is "an effective and equal medium for exposure therapy."

A second 2019 meta-analysis, by Wechsler, Kümpers and Mühlberger, asked a narrower and harder question: in head-to-head trials only, does VR match in-person exposure? Across 9 RCTs and 371 participants, the overall difference favored in-vivo exposure slightly but not significantly (g = −0.20). Broken out by condition, specific phobia showed no meaningful gap and agoraphobia showed essentially none — but social phobia showed a medium, statistically significant advantage for in-person exposure (g = −0.50). That is a genuinely mixed result, and it points at something sensible: a simulated crowd is a harder thing to get right than a simulated height.

A larger and more recent pooled analysis, published in Frontiers in Psychiatry in 2025, gathered 33 trials and 3,182 adolescents and adults with anxiety disorders and reported a large benefit for VR-based intervention over conventional care (SMD = −0.95). The same authors flagged heterogeneity of 91 percent, mostly high or unclear risk of bias in the included studies, and thin long-term follow-up data. A big number with a big asterisk is still worth knowing about, but it is not the same as a settled one.

Can it work without a therapist in the room?

One of the more striking trials came out of Oxford in 2018, when Freeman and colleagues tested a fully automated VR treatment for fear of heights in The Lancet Psychiatry. There was no human therapist. A virtual coach guided 100 participants — people who had lived with the fear for an average of 30 years — through the program. Ninety percent completed the full course, and the between-groups effect size on the primary outcome was d = 2.0, maintained at follow-up.

That study is often quoted as proof that software can replace clinicians, which overstates it. Fear of heights is one of the tidiest phobias to simulate and one of the most responsive to exposure. What the trial does establish is a real proof of concept: for at least one well-defined fear, the therapeutic content can be delivered by the program itself.

What about fear of needles and blood?

This is the corner of the literature closest to our own interest, and it is also the thinnest.

Jiang, Upton and Newby ran a randomized wait-list controlled pilot of single-session VRET for blood-injection-injury phobias, published in the Journal of Affective Disorders in 2020. At three-month follow-up, they reported moderate to large improvements on injection and injury fear measures, with Hedges' g values from roughly 0.63 to 1.14, and 19 of 20 participants reporting more confidence about facing real-world blood and needle situations. The authors' own reading was measured: one session may be a useful adjunct or an interim step before in-vivo exposure rather than a standalone treatment.

A 2025 pilot in Frontiers in Psychiatry by Caltabiano and colleagues went further into the mixed column. Sixty-two adults with needle fear, 32 of them doing two self-paced virtual needle-procedure exposures. The exposures reliably provoked anxiety, which is the mechanism working as intended. At one-month follow-up, one measure showed a modest but statistically significant drop in symptom severity while another showed no meaningful change at all, and some participants reported VR side effects. Usability ratings were high. Certainty was not.

Blood-injection-injury fear also has a quirk that sets it apart from other phobias: it is the one commonly associated with fainting rather than flight, which is why clinics handle it differently from ordinary nervousness. Whether exposure protocols designed for spiders and balconies transfer cleanly to it is still an open question.

Where the evidence is thinner than the headlines

Three honest caveats are worth carrying around.

Sample sizes are small. Carl's entire meta-analysis rests on about a thousand people across four diagnostic categories, and the head-to-head literature on roughly 371. Meta-regression in that same analysis found larger trials producing smaller effects, which is the classic signature of early enthusiasm cooling as studies get more rigorous.

Follow-up is short. Most trials check in at one to three months. Phobias are measured in decades — the Oxford participants had held theirs for 30 years — so a three-month window tells you about the beginning of a change, not its durability.

And the numbers floating around the wider internet are worse than the ones in the journals. "Success rates of 66 to 90 percent" circulates widely without a primary source attached. The published effect sizes are more modest and more interesting.

None of this makes VR exposure therapy a weak bet. Across two decades it has been tested more thoroughly than most digital health tools ever are, and it keeps landing in the same place: real effect, comparable to the in-person alternative for simple phobias, less convincing for social ones, under-studied for needles.

A distinction worth keeping straight

Because it comes up constantly: SendorVR is not an exposure therapy product and does not deliver the treatment described above. It is a non-clinical comfort and engagement aid — a short activity for a short moment, designed to give attention somewhere pleasant to go during a routine appointment. The research summarized here is about a different use of the same hardware, conducted by other people, with other goals. We find it worth reading precisely because it is not about us. You can find more of what we read on our science page.

Frequently asked questions

Is VR exposure therapy as good as real-world exposure?

For specific phobias, pooled trial data suggests it is roughly comparable, with no statistically significant difference in the largest meta-analysis. For social anxiety, head-to-head evidence has favored in-person exposure by a medium margin. It depends heavily on which fear is being treated.

Do the effects last?

Where follow-up exists, gains have generally held at one to three months, and the Oxford heights trial maintained its effect at follow-up. Long-term data beyond a few months is scarce across the literature, which the researchers themselves consistently flag.

Has VR been studied specifically for needle fear?

Yes, but only in small pilots so far. A 2020 wait-list controlled pilot and a 2025 pilot both reported some improvement on some measures, with mixed results across others and calls from both author groups for larger trials.

Is a VR headset used during a procedure the same thing as exposure therapy?

No. Exposure therapy involves repeated, graded confrontation with the feared thing, with the goal of lasting fear reduction over a course of treatment. A headset used during an appointment is doing something different, and is studied in a separate body of research — one that has more to say about how a procedure feels in the moment than about long-term fear.

Are there downsides to VR exposure?

Reported side effects in trials include nausea, disorientation and eye strain, and some participants in the 2025 needle pilot experienced them. Simulation quality also varies, and a poorly realized environment may not provoke enough of a response for exposure to do its work.

This article discusses published research on virtual reality generally. SendorVR has not been evaluated in these studies, and no product-specific claims are made or implied. SendorVR is a non-clinical comfort and engagement aid; all medical care remains with qualified healthcare professionals.

Sources

See How SendorVR Works in Your Clinic

Book a short demo with our team to see how SendorVR can be deployed in your environment.