Honey on Skin: What the Research Actually Shows

The strongest thing honey has done to skin in a trial happened on wounds, not faces: in Cochrane’s 2015 systematic review, dressings of sterilised medical-grade honey healed partial-thickness burns about 4.7 days faster than conventional treatment, on high-quality evidence. On facial skin the record splits three ways, and all three randomised trials used the same medical-grade kanuka honey product: negative for acne, positive for rosacea, and no difference from the antiviral cream it was measured against for cold sores — a null result, not a demonstration that the two are equivalent. None of the three tested honey from a jar, and none asked a cosmetic question: no controlled trial of honey applied to healthy skin for a cosmetic outcome was located for this page, and on hair there is no controlled trial at all. What honey verifiably does in a cream is chemistry — it binds water — and each of these findings repays a closer look.

Everything below is the published literature, cited as it comes.

Plate I · ArtefactRecord

Why does the wound research not carry over to a face?

Because three things change at once — material, setting and tissue — and each changes what the finding means.

Start with the finding itself. The anchor evidence is a Cochrane systematic review (Jull and colleagues, 2015): 26 randomised trials, 3,011 participants. Honey dressings healed partial-thickness burns about 4.7 days faster than conventional treatment — which in the pooled trials included polyurethane film, paraffin gauze, soframycin-impregnated gauze, sterile linen and leaving the burns exposed (2 trials, 992 people, high-quality evidence — the review’s top grade, meaning further research is unlikely to shift the result). Honey also healed infected post-operative wounds faster than antiseptic washes plus gauze in one trial of 50 people, at moderate-quality evidence — a grade meaning further research well might shift it. The comparison honey lost is just as instructive: early excision and grafting healed partial and full thickness burns about 13.6 days faster than honey (1 trial, 50 people, low-quality evidence). Beyond the burns and post-operative comparisons — leg ulcers, diabetic foot ulcers, minor and mixed chronic wounds — the reviewers rated the evidence low or very low quality, and said it “does not form a robust basis for decision making”.

The material in those trials is a manufactured medical product, not food. The review describing one of the first medically certified examples (Simon and colleagues, 2009) records it as licensed for professional wound care in Europe and Australia, and answers the sterility question directly: “Why is medical honey irradiated? Honey may contain spores from Clostridium botulinum, which are inactivated. The ready to use product is delivered sterile.” Food honey has had no such step.

And a wound is broken skin repairing itself, dressed and monitored — a cosmetic complaint on intact skin is a question these trials never asked. The full wound record is on the wounds page, and the laboratory story behind graded manuka on the manuka page.

Does honey do anything for acne?

One randomised controlled trial has asked, and it came back negative.

Semprini and colleagues (BMJ Open, 2016) recruited 136 people aged 16 to 40 with diagnosed acne in New Zealand. Everyone washed twice daily with a triclocarban antibacterial soap for 12 weeks; the honey arm also applied 90% medical-grade kanuka honey with 10% glycerine. The primary outcome — a fall of at least 2 points on the Investigator’s Global Assessment — was reached by 4 of 53 on honey (7.6%) and 1 of 53 controls (1.9%): odds ratio 4.2, 95% confidence interval 0.5 to 39.3, p = 0.17. A confidence interval that wide, straddling 1, is the statistics saying the trial cannot tell a real effect from none. The authors’ summary: “The results presented do not support a benefit of using honey in addition to a common over-the-counter antibacterial soap.”

The trial’s limits are worth learning from, because they show how a skin trial can be blunted. Participants could not be blinded — the product smelt of honey. Withdrawals ran to about a fifth of each arm. And a floor effect weakened the measure: roughly a third of participants entered at a score of 2 and could not improve by 2 points without clearing entirely. HoneyLab funded the study and supplied the product. The authors were “not aware of an RCT considering a honey-based therapy for acne” — this was the first, and a search found no later one, only laboratory studies of honey against acne-associated bacteria in culture. The honey tested was kanuka, not manuka; no trial of manuka honey for acne was located.

Fig. I · The acne trial, countedDiagram

Semprini and colleagues · BMJ Open, 2016

Randomised controlled trial. 136 people aged 16 to 40 with diagnosed acne, in New Zealand, over 12 weeks.

Primary outcomeA fall of at least 2 points on the Investigator’s Global Assessment.

Honey arm

Twice-daily triclocarban soap wash, plus 90% medical-grade kanuka honey with 10% glycerine.

4 of 53 reached it (7.6%)— 49 did not

Control arm

Twice-daily triclocarban soap wash alone.

1 of 53 reached it (1.9%)— 52 did not

Reached the primary outcomeDid not reach it

Odds ratio and 95% confidence interval · axis spaced logarithmically

4.2

0.5139.3

The dashed line stands at an odds ratio of 1 — no difference between the arms. The interval crosses it: it runs from 0.5 to 39.3, with p = 0.17. The authors reported no evidence of benefit.

What limits it — the authors’ own list

• Participants could not be blinded, because the product smelt of honey.
• Withdrawals ran to about a fifth of each arm.
• A floor effect blunted the measure: roughly a third entered at a score of 2 and could not improve by 2 points without clearing entirely.
• HoneyLab funded the study and supplied the product.
• The honey tested was kanuka, not manuka — no trial of manuka honey for acne was located.

The same team and product produced a positive result in a different condition, and it deserves the same close reading. In an 8-week trial of 138 adults with rosacea (Braithwaite and colleagues, 2015), 24 of 68 on honey and 12 of 69 on a control cream improved by at least 2 points on the investigator’s severity score: relative risk 2.03 (95% CI 1.11 to 3.72, p = 0.020) — twice the rate of improvement, in an interval that stays above 1. Participants could not be blinded, though the assessing investigator was; the trial was manufacturer-funded, no replication was located, and the product was again a registered medical-grade preparation, not honey from a cupboard.

A third trial of the same product took on a drug rather than a cream, and taught a different lesson. Semprini and colleagues (BMJ Open, 2019) randomised 952 adults with herpes simplex labialis, recruited through 76 New Zealand community pharmacies, to the same 90% kanuka honey with 10% glycerine or to 5% aciclovir. Median time to return to normal skin was 9 days on honey and 8 on aciclovir: hazard ratio 1.06, 95% confidence interval 0.92 to 1.22, p = 0.56. The authors reported “no evidence of a difference in efficacy” — which, in a trial designed to detect superiority and run without a placebo arm, is a null result and not a demonstration that the two are equivalent: a trial built to find a winner and finding none cannot certify a tie. It was open-label, the honey could not be masked, lesions were not assessed by a clinician, and Honeylab funded it.

If a skin condition is why you are reading this, a pharmacist, GP or dermatologist is the person to see — they can look at the skin in question, which no trial summary can.

What is honey doing in a cosmetic product?

Holding water. Honey is a viscous supersaturated sugar solution, and concentrated sugar binds water — which in formulation vocabulary makes it a humectant: a property of the mixture, not an effect on a person.

A narrative review in the Journal of Cosmetic Dermatology (Burlando & Cornara, 2013) describes the practice: honey at 1 to 10% in cosmetic formulations, reaching “up to 70%” when mixed with oils, gels and emulsifiers, across creams, lotions, shampoos and conditioners. It lists emollient, soothing and hair-conditioning effects — but it reviews a field, not controlled trials, and that distinction is worth carrying everywhere on this subject. No controlled trial of honey on healthy skin for a cosmetic outcome was located; an ingredient having a function in a cream is not evidence that a face looks different for it.

Is honey good for your hair?

Not on any evidence this review could locate. No controlled trial of honey applied to hair was located; the familiar claims — shine, growth, conditioning — rest on nothing published that this review could find. That emptiness is itself the finding.

The nearest study is not about hair, and it is a good specimen of the weakest study design still called clinical. Al-Waili (2001, European Journal of Medical Research) treated 30 people aged 15 to 60 with chronic seborrhoeic dermatitis of the scalp, face and chest, using a 90% crude-honey-in-water preparation applied over four weeks. All improved on unblinded investigator assessment — the lesions were examined — and the patients reported subjective improvement in hair loss. Those who improved then entered a six-month follow-on phase in which half continued weekly honey and half served as controls: none of the 15 on honey relapsed, against 12 of 15 untreated. Read as design, though: no control arm and no randomisation in the four-week treatment phase, no randomisation described for the split that followed, no blinding at any point, 30 people in total, and no replication located in the twenty-five years since. Without randomisation or blinding, a study cannot separate the substance from expectation, selection and time — whichever direction its numbers point.

What are the real risks worth knowing?

Two, and neither is theoretical.

The first is allergy. Reactions to honey are rare but documented, up to anaphylaxis, usually traced to pollens carried in the honey or to bee proteins such as Major Royal Jelly Protein 1 (Matuszewska-Mach and colleagues, 2025). Contact reactions are better documented: propolis, the bee resin found in traces in honey and beeswax, is a recognised contact allergen, and the same review cites patch-test data in which 6.2% of 257 dermatitis patients in Italy reacted to it.

The second is sterility. Honey sold as food is not sterile — which is why the wound-care product is irradiated, and why unsterilised honey does not belong on broken skin.

How have regulators read the evidence?

Under three separate rulebooks, and the differences are instructive.

As food, health claims are governed by the Great Britain nutrition and health claims register: only authorised claims may be used commercially. Every bee product entry — honey, propolis, royal jelly, honeycomb — is marked non-authorised. Skin appears there only under royal jelly: the claim to a “beneficial effect on skin” (EFSA Journal 2011;9(4):2083), and, under lyophilised royal jelly, “overall rejuvenation of the skin and human body” (EFSA Journal 2010;8(10):1738 and 2010;8(10):1799) — all assessed, all refused. Honey’s own eight entries are all non-authorised, none of them about skin. Reading those refusals is a shortcut to the state of the evidence: the assessors looked at the same literature this page does and found it insufficient.

Cosmetics sit under a different regime: Regulation (EC) No 1223/2009, as amended for Great Britain in 2019. Labelling and advertising “must not imply they have characteristics or functions which they do not have”, and the responsible person must “be able to prove that any claims made when marketing the product are correct”. Cosmetic claims never appear on the health claims register, because they are not health claims.

Wound care is a third regime again: those dressings are certified medical products, covered on the wounds page.

Fig. II · One substance, three rulebooksDiagram

One substance — honey — read under three regimes

As food

The rulebook

The Great Britain nutrition and health claims register.

What may be said

Only authorised claims may be used commercially.

Where honey stands

Every bee product entry — honey, propolis, royal jelly, honeycomb — is marked non-authorised. Honey’s own seven entries are all non-authorised, none of them about skin.

Skin appears on the register only under royal jelly: “beneficial effect on skin” (EFSA Journal 2011;9(4):2083) and, under lyophilised royal jelly,“overall rejuvenation of the skin and human body” (EFSA Journal 2010;8(10):1738 and 2010;8(10):1799) — all non-authorised.

Authorised skin claim available to a seller of honey — none

As a cosmetic

The rulebook

Regulation (EC) No 1223/2009, as amended for Great Britain in 2019.

What may be said

Labelling and advertising “must not imply they have characteristics or functions which they do not have”, and the responsible person must“be able to prove that any claims made when marketing the product are correct”.

Where honey stands

Cosmetic claims never appear on the health claims register, because they are not health claims.

This house sells no cosmetic, and nothing here is one.

As a wound dressing

The rulebook

Certified medical products.

What may be said

Not drawn — this page states no claims rule for this regime. What it records is a licence: one of the first medically certified examples is recorded as licensed for professional wound care in Europe and Australia.

Where honey stands

A manufactured medical product, not food. The ready to use product is delivered sterile; food honey has had no such step.

A wound is broken skin repairing itself, dressed and monitored — a cosmetic complaint on intact skin is a question these trials never asked.

Three regimes, and they share no rulebook.

No controlled trial of honey on healthy skin for a cosmetic outcome was located for this page.

So what does the record actually show?

A clear shape, once the borrowed confidence is set aside. The clinical evidence belongs to sterilised products used on wounds, and is mixed even there — the same review that credits honey dressings with faster burn healing also records early excision and grafting healing burns about 13.6 days faster than honey. Three trials have reached facial skin, all of one manufactured product, all funded by its manufacturer, none independently replicated: negative for acne, positive for rosacea, and null against an antiviral cream for cold sores. The cosmetic story is a formulation property — honey binds water — rather than a measured outcome; the hair story has never met a controlled trial; and the best-documented chapter is the risk side: allergy, and the sterility gap between a food jar and an irradiated dressing. On this subject, the closer the reading, the more specific the facts become — which is exactly what makes it worth reading closely.