The research on NLP is thin, and where it exists it splits cleanly in two. The distinctive claims that made Neuro-Linguistic Programming famous, above all preferred representational systems and eye movements as a read on internal processing, have been tested since the early 1980s and have not been supported. The ordinary communication skills at its core, rapport, precise questioning, deliberate state change and outcome setting, do have supporting evidence, but almost all of it comes from psychology outside NLP rather than from studies of NLP itself.
That distinction is what most pages on this topic skip. Below are the main studies and reviews by name, what each one looked at, and what it found, so you can check the claims yourself rather than take anyone’s summary on trust.
| Study or review | Year | What it examined | Finding |
|---|---|---|---|
| Sharpley, Journal of Counseling Psychology | 1984 and 1987 | Preferred representational systems and predicate matching | No reliable support |
| Heap, review chapter in Hypnosis | 1988 | Around 63 studies on core NLP claims | Claims not substantiated |
| Witkowski, Polish Psychological Bulletin | 2010 | 315 articles, 63 empirical studies in indexed journals | Roughly 18 percent supportive, 55 percent non-supportive |
| Sturt and colleagues, British Journal of General Practice | 2012 | Systematic review of NLP effects on health outcomes | Little evidence of effect, weak study quality |
| Wiseman and colleagues, PLoS ONE | 2012 | Eye movements as an indicator of lying | No relationship found |
| Zaharia, Reiner and Schutz, Psychiatria Danubina | 2015 | Meta-analysis of NLP-based psychotherapy | Small positive effect, small samples, authors inside the field |
What does the research on NLP actually say?
That the theory failed and parts of the practice were never properly tested. Those are different verdicts and they get collapsed together constantly, by supporters and by critics.
The most cited single source is Tomasz Witkowski’s 2010 survey in the Polish Psychological Bulletin. He assembled a database of 315 articles on NLP and found 63 empirical studies published in journals with a recognised impact factor. Of those, he classified about 18 percent as supportive, a quarter as uncertain and 55 percent as non-supportive. His conclusion was that NLP functions as what he called pseudoscientific decoration.
The other reference point for anyone in a health or coaching setting is the systematic review by Jacqueline Sturt and colleagues, published in the British Journal of General Practice in 2012. They looked for randomised and controlled evidence that NLP improves health outcomes, found ten studies worth including, and rated the methodological quality as poor throughout. Their conclusion was that there is little evidence NLP works for health-related problems, not that it had been shown to fail, which is a weaker and more accurate statement.
Which NLP claims have been tested and failed?
Two in particular, and they are the two most people have heard of. Both were tested early, repeatedly, and by researchers with no stake in the outcome.
- Preferred representational systems. The claim that each person has a dominant sensory channel, detectable from their verbs, and that matching it improves rapport and results. Christopher Sharpley reviewed the literature in the Journal of Counseling Psychology in 1984 and again in 1987 and found the data did not support it.
- Eye accessing cues. The claim that eye direction reliably indicates whether someone is picturing, hearing or feeling, and by extension whether they are recalling or constructing. Richard Wiseman and colleagues tested the lie detection version of this across three studies in PLoS ONE in 2012 and found no relationship at all.
Both are still taught in some trainings, which is the single clearest signal of a provider who has not read the literature. Our glossary entry on eye accessing cues and what they do and do not show sets out the original claim and the results side by side.
Which parts have evidence behind them?
The generic skills, and the support comes from mainstream psychology rather than from NLP research. This is an awkward point for the field because it means the useful parts are not original, and it is the honest position.
Behavioural mirroring is the clearest example. Tanya Chartrand and John Bargh’s 1999 chameleon effect studies in the Journal of Personality and Social Psychology showed that people unconsciously copy each other’s mannerisms and that being copied increases liking and smoothness of interaction. That is the mechanism behind matching in NLP, established independently and long after NLP started teaching it.
Specific and difficult goal setting has a similar story. Edwin Locke and Gary Latham’s decades of goal setting research support the core of what NLP calls a well formed outcome: define it precisely, state it in the positive, make it something you can verify. Mental rehearsal of a performance, which NLP calls future pacing, has its own literature in sport psychology. If you want the wider frame for what NLP claims to be, our page on what NLP is and what it covers lays out the model that these skills sit inside.
Why is there so little good NLP research?
Four reasons, and none of them is a conspiracy against the field.
- No standard protocol. Two trainers using the same technique name may do noticeably different things, which makes replication difficult and results hard to interpret.
- No research funding. NLP developed as a commercial training field. Nobody is paid to run trials, and the people best placed to run them sell the training.
- Conflicted authorship. The most positive result, the 2015 meta-analysis in Psychiatria Danubina reporting a moderate effect for NLP-based psychotherapy, was produced by authors working inside the field, drawing on small and mostly uncontrolled studies.
- Terminology drift. The vocabulary has expanded for forty years without a governing body, so a claim tested in 1985 may not describe what a trainer means by the same word today.
The result is a field where the burden of proof has not been met, rather than one that has been disproved wholesale. Those are not the same thing, and the difference matters if you are deciding what to learn.
How should you read a provider’s evidence claims?
Treat any page that says NLP is scientifically proven as a warning. The literature does not support that sentence, and a trainer who writes it either has not read it or is counting on you not to. Five checks separate a careful provider from a careless one.
- Ask which claims they consider unsupported. Anyone who says none is not worth your week.
- Look for named studies with dates, not the phrase studies show.
- Check whether eye accessing cues and sensory preference are taught as fact or as history.
- Ask what the training does not do. Honest answers exist, and they include serious mental health conditions.
- See whether the claimed outcomes are behavioural, such as asking better questions, or magical, such as reprogramming your subconscious in a weekend.
I teach the failed claims as part of the course rather than leaving them out, because a practitioner who cannot say which parts are evidence poor will eventually be asked by a client and will have nothing to say. That framing runs through the small group Practitioner training we run in English in Lisbon, and it makes the honest parts easier to use, not harder. The related question of whether the field as a whole counts as proven is worked through in our piece asking whether NLP is scientifically proven.
NLP is a set of communication and change tools. It complements professional medical or psychological care and does not replace it.
Carolin Mallmann, Licensed Trainer of The Society of NLP, Lisbon
Frequently asked questions
Is there any research evidence for NLP?
There is research, and most of it is negative on NLP’s distinctive claims. Witkowski’s 2010 survey found that of 63 empirical studies in indexed journals, roughly 18 percent supported NLP claims and 55 percent did not. The generic communication skills NLP teaches have support, but from mainstream psychology rather than from NLP studies.
Which study is the most cited against NLP?
Tomasz Witkowski’s 2010 article in the Polish Psychological Bulletin, which analysed a database of 315 NLP articles and classified the empirical studies by outcome. For health settings specifically, the 2012 systematic review by Sturt and colleagues in the British Journal of General Practice is the standard reference.
Have eye accessing cues been disproved?
The lie detection version has. Wiseman and colleagues ran three studies published in PLoS ONE in 2012 and found no relationship between eye direction and deception. Earlier work through the 1980s also failed to support eye movements as a reliable indicator of which sensory channel someone is using.
Is any NLP meta-analysis positive?
One is. A 2015 meta-analysis in Psychiatria Danubina reported a small to moderate positive effect for NLP-based psychotherapy. It should be read with care: the authors work inside the field, and the included studies were small and largely without control groups.
Why has NLP not been researched properly?
There is no standard protocol to test, no research funding in a commercial training field, and no governing body keeping the terminology stable. The people best placed to run trials are the people selling the courses, which creates an obvious conflict and few incentives to publish negative findings.
Does weak evidence mean NLP is useless?
No, it means the claims are unproven rather than proven false, and that the burden of proof sits with the field. Skills such as building rapport, asking precise questions and setting clear outcomes are useful whether or not the NLP theory around them holds. Treat any promise of guaranteed results as marketing.
Keep exploring
- Is NLP scientifically proven?, for the same question answered at the level of the field rather than study by study.
- What did Bandler and Grinder create?, for which of the original models the research has treated kindly and which it has not.
Carolin Mallmann is a Licensed Trainer of The Society of NLP, trained by Richard Bandler. She has trained more than 300 people and runs English language NLP trainings in Lisbon in groups of no more than 10.
