Is Honey Good for You? What the Research Says

Honey is sugar with character — about 82% sugars by weight on the standard reference analysis — and one of its uses has genuinely good trial evidence behind it: for a cough in a child over one, honey probably works for the first few days, and Cochrane’s reviewers found it probably beats placebo and no treatment. This page walks through what has actually been measured — the cough trials in full, the metabolic head-to-heads, the hay-fever experiments, and how Europe’s food regulator read the wider evidence.

Everything below is the published literature, cited as it comes; the rest of the evidence section reads the neighbouring shelves the same way.

Plate I · IllustrationRecord

Fig. I · The evidence-strength ledgerDiagram

Popular uses of honey, by strongest evidence located

1. Authorised GB claim
   No authorised health claim available to makeEight honey entries in the GB register — every one non-authorised
2. Systematic review of trials
   Eases a cough, soothes the throatNo strong evidence for or against — about three days
3. Single human trials
   Healthier than sugar; low-GIMetabolically indistinguishable from sucrose — measured GI 69–74

   Natural defences; aids digestionMeasured in two small trials; no functional effect found

   Local honey relieves hay feverThree small trials, results conflicting — none tested honey local to the patient
4. Laboratory dish
   Eating it fights infectionDishes and dressings, not eating honey
5. Nothing located
   No use drawn here sits at this rungSome human trial evidence was located for every use above — mostly small, and mostly null

What is honey, nutritionally?

Sugar, water, and very little else. The standard reference analysis (USDA FoodData Central) puts honey at 82.4 g of carbohydrate per 100 g, of which 82.1 g is sugars — mostly fructose (about 41 g) and glucose (about 36 g) — with 17.1 g of water, 304 kcal, 0.3 g of protein and no fat. The minerals sit at trace level: 6 mg of calcium and 0.42 mg of iron per 100 g. The shelf line “rich in vitamins and minerals” fails on arithmetic.

UK dietary guidance classifies honey accordingly. The NHS counts the sugars in honey as free sugars — the same category as table sugar and syrups — and names free sugars as the kind adults should cut down on: no more than 30 g a day for adults, 24 g for children aged 7 to 10, 19 g for ages 4 to 6. A typical tablespoon of honey weighs about 21 g and carries about 17 g of sugars — more than half the adult daily allowance on one spoon. The same guidance names sugar as one of the main causes of tooth decay; honey sits inside that sentence, not beside it.

Plate II · IllustrationRecord

Does honey help a cough?

Probably — for the first few days, in a child over one. This is the best-tested thing on the page, and it rewards reading in full.

The anchor is a Cochrane systematic review (Oduwole and colleagues, 2018 update): six randomised controlled trials, 899 children, honey against dextromethorphan, diphenhydramine, salbutamol, bromelin, placebo and no treatment. Its central finding, verbatim: “Honey probably relieves cough symptoms to a greater extent than no treatment, diphenhydramine, and placebo, but may make little or no difference compared to dextromethorphan.” That “probably” is a term of art. The certainty was rated moderate against placebo and no treatment, and in Cochrane’s grading moderate certainty means the reviewers think the true effect is likely close to what the trials measured, while leaving real room for further trials to shift it. The effect was also time-boxed: up to three days, honey was probably more effective than placebo or salbutamol; beyond that, probably no advantage — the shape of a remedy for the short, self-limiting cough of a cold. The limitations sit inside the review itself, stated once and worth knowing: most trial children were treated for a single night; the reviewers judged two studies at high risk of performance and detection bias and three at unclear risk of attrition bias; and of the six trials, two were funded by honey industry bodies (the Honey Board of Israel and the USA National Honey Board), two by pharmaceutical manufacturers, and one reported no funding source. On safety, the reviewers conclude “There was no difference in occurrence of adverse events between the honey and control arms”, with two figures behind that sentence: gastrointestinal symptoms against placebo pooled at RR 1.91 (95% CI 1.12 to 3.24; 2 studies; 402 children; moderate certainty), and 9.3% of children on honey reporting nervousness, insomnia or hyperactivity against 2.7% on dextromethorphan (RR 2.94, 95% CI 0.74 to 11.71; low certainty). The review excluded infants under 12 months, for whom honey is not advised at all (see safety, below). The reviewers’ overall summary — “There was no strong evidence for or against using honey” — sits beside the moderate-certainty findings without contradiction: read together, the two statements say the short-term comparisons held up at moderate certainty inside a small evidence base.

A broader systematic review and meta-analysis (Abuelgasim and colleagues, BMJ Evidence-Based Medicine, 2021) pooled 14 studies of honey for upper respiratory tract symptoms and found honey superior to “usual care” for symptom relief: a combined symptom-score difference of −3.96 (95% CI −5.42 to −2.51), with cough frequency and severity also favouring honey (SMD −0.36 and −0.44). The teaching point is in the comparator. “Usual care” mostly meant over-the-counter remedies the authors themselves classed as ineffective — antihistamines, cough syrups, guaifenesin, salbutamol — so beating it is a lower bar than beating placebo. Against placebo the picture is genuinely unsettled: only two placebo-controlled trials used comparable scales and could be pooled, and they disagreed sharply — Cohen 2012 favoured honey (SMD −1.03, 95% CI −1.32 to −0.75), Canciani 2014 found nothing (SMD −0.20, 95% CI −0.59 to 0.19) — and pooling them gave SMD −0.63 (95% CI −1.44 to 0.18) with heterogeneity of 91%, a statistician’s way of saying the two trials disagree so thoroughly that their average settles little. A third placebo-controlled trial (Waris 2014) could not be pooled and reported honey ahead of placebo by the final day (mean difference 3.99, p = 0.003). The authors rate the overall risk of bias moderate and call for further high-quality placebo-controlled trials.

That is where honey’s clinical evidence peaks: for a cough in a child over one, honey probably works for the first few days. Neither review carries the evidence past the short-lived symptoms of an upper respiratory infection.

What has been measured for the other claims?

The ledger, claim by claim:

Three rows deserve unpacking.

The antibacterial row is the great category confusion of the honey shelf, and worth understanding properly. The literature behind it is real — medical-grade honeys show potent bactericidal activity against antibiotic-resistant bacteria in laboratory assays, and clinical observations exist for honey applied to wounds (Mandal & Mandal, 2011) — but it is a literature of dishes and dressings, not of eating honey. An in-vitro result is a different kind of fact from a trial: it describes what a honey does to bacteria in culture, at dish concentrations, and says nothing by itself about what swallowing honey does in a person. The same laboratory record powers graded manuka; the manuka evidence page reads it on its own terms.

The blood-sugar rows carry the cleanest data on this page, thanks to an unusually strict design. In a crossover trial, every participant consumes each sweetener in turn, so each person serves as their own control and differences between people cancel out — a sharp instrument for a nutrition question. Raatz and colleagues (2015) fed 55 adults — 28 glucose-tolerant, 27 glucose-intolerant — 50 g of carbohydrate a day from honey, sucrose and high-fructose corn syrup, two weeks each, and found the three metabolically indistinguishable: glucose, insulin, insulin-resistance and glucose-tolerance measures did not differ, and triglycerides rose on all three. The measured glycaemic index of four honey varieties (Ischayek & Kern, 2006; 12 adults, crossover) ran 69 to 74 against a glucose reference — none of them low, with no significant difference between varieties and no relationship to the fructose-to-glucose ratio. The wider pooled literature on honey and blood lipids is heterogeneous and low-certainty, and does not settle the question either way.

The hay fever row is the one most often reported backwards. The folk claim — that honey from nearby hives carries the local pollens that provoke the symptoms — has been put to a direct test once, and it failed it. Rajan and colleagues (2002) randomised 36 adults with allergic rhinoconjunctivitis, recruited at a single university health centre, to one of three arms, the first of them “locally collected, unpasteurized, unfiltered honey”; a tablespoon a day, against corn syrup with synthetic honey flavouring. Their reported result: “Neither honey group experienced relief from their symptoms in excess of that seen in the placebo group.” The other two trials were not built around locality at all. Asha’ari and colleagues (2013) added Tualang honey at 1 g/kg of body weight daily to loratadine in 40 adults and reported improvement in individual symptoms — but those are within-group paired comparisons inside each arm separately (t tests at df 19), and the trial’s only between-group comparison of symptom scores shows no significant difference at any timepoint, p = 0.428 at week 8. Saarinen and colleagues (2011) gave 44 patients honey deliberately spiked with birch pollen or plain honey; the 60% lower symptom score is against a separate, unblinded medication-only group, and the randomised comparison, spiked honey against plain honey, showed no significant symptom difference. The hay fever page reads all three line by line.

How regulators judged the evidence

The Great Britain nutrition and health claims register is the authoritative list for England, Scotland and Wales: only claims authorised in it may be used in commercial communications about food. For honey it contains eight entries, and every one is non-authorised. The rejected wordings are the shelf’s vocabulary verbatim: “Helps maintain your natural defences”, “natural antimicrobial action”, “Helps support and maintain a healthy digestive system”, antioxidant-capacity claims, honeycomb claims for memory and menopause.

The reasons are more interesting than the verdicts. Most claims failed at the first hurdle: the underlying EFSA assessments (EFSA Journal 2011;9(6):2243 and 2010;8(2):1484, cited row by row in the register) concluded that honey “is not sufficiently characterised for a scientific assessment of this claimed effect” — honey varies too much, jar to jar, for any claimed effect to be pinned to the food at all. The antioxidant claims went a step further and were rejected on the merits: assessed against protection of DNA, proteins and lipids from oxidative damage (EFSA Journal 2010;8(2):1489), the claimed effect “has not been substantiated”.

One juxtaposition teaches more than either fact alone: “soothing for the throat” — the closest wording to the one use with respectable trial evidence — is itself non-authorised. That is not the register contradicting Cochrane; the two apply different tests. Cochrane asks whether trials show an effect; the register asks whether the effect is proven well enough, in a food consistent enough, to print on a label — a deliberately higher bar.

Fig. II · The register’s honey rowsDiagram

Great Britain nutrition and health claims registerThe authoritative list for England, Scotland and Wales: only claims authorised in it may be used in commercial communications about food.Honey entries 8

Status

“Soothing for the throat”The closest wording to the one use with respectable trial evidence — itself a rejected claim.

Non-authorised

“Helps maintain your natural defences”

Non-authorised

“Natural antimicrobial action”

Non-authorised

“Helps support and maintain a healthy digestive system”

Non-authorised

Claims of protection of DNA, proteins and lipids from oxidative damageThe one line the page says went further — assessed on the merits, not turned back at characterisation. Written in the plural on this page.

Non-authorised

Honeycomb claims for memory and menopauseWritten in the plural on this page.

Non-authorised

In Great Britain, a seller of honey has no authorised health claim available to make.Six lines are named on this page; two of them are written in the plural, so whether they account for all eight entries or leave a balance is not stated here.

None authorised

Grounds recorded on this page

Most failed at the first hurdleHoney “is not sufficiently characterised for a scientific assessment of this claimed effect”.EFSA Journal 2011;9(6):2243 and 2010;8(2):1484, cited row by row in the register. Which rows the “most” covers is not stated, so no row above carries this ground.

One line went further — assessed on the meritsAssessed against protection of DNA, proteins and lipids from oxidative damage, the claimed effect “has not been substantiated”.EFSA Journal 2010;8(2):1489.

Rows in quotation marks are the register wordings this page quotes; rows without them are described on the page, not quoted.

Does raw honey have more going for it than processed?

Not on any trial reviewed here, and only faintly on one pooled analysis. “Raw” is a claim about handling — honey that has not been pasteurised or fine-filtered after extraction; what the word means, and how a raw claim can be evidenced, is set out in the guide to what raw honey is and in raw versus pasteurised. Those differences are real, and they matter for character and quality; none of the individual trials above turns on them. The cough trials did not compare raw with processed; the metabolic trial set honey against sucrose and corn syrup; the glycaemic-index trial found even varietal differences made no measurable difference to blood-glucose response. And the register’s rejections apply to honey as a category, however handled. One meta-analysis has looked directly: pooling 18 controlled trials, Ahmed and colleagues (2023) report subgroup differences by processing, with raw honey among the subgroups showing effects on fasting glucose and total cholesterol — at low certainty of evidence, from subgroup analysis. Subgroup findings are hypothesis-generating: they point to the trial someone should run next, rather than delivering a result to rely on. On the current record, the raw-versus-processed question has been asked once, at low certainty, and not yet answered convincingly.

Who should be careful with honey?

Two lines in the record are firm.

The first is absolute: no honey for babies under 12 months. NHS guidance is that honey occasionally contains bacteria that can produce toxins in a baby’s intestines, leading to infant botulism — a very serious illness. The rule covers every honey, however handled or graded. The same NHS page adds, unprompted, the theme of this whole article: “Honey is a sugar, so avoiding it will also help prevent tooth decay.”

The second is the allergy record. Allergy to eaten honey is rare but documented — including anaphylaxis — in children and adults. A review of the case literature identifies the culprits as Major Royal Jelly Protein 1 (MRJP1), the dominant protein in both royal jelly and honey, and plant pollens carried in the honey. Documented cases run from generalised urticaria with breathing difficulty in a five-year-old to intensive-care-level anaphylaxis in an adult; no population prevalence figure exists in the source. The record: rare, case-report level, and real.

For everyone else the safety question is the sugar question: honey spends against the NHS free-sugar allowance exactly as table sugar does.

So — is honey good for you?

Read as a whole, the record says this: honey is a distinctive food that is four-fifths sugar, and UK guidance treats it as sugar. One genuine island of clinical evidence exists — short-term cough relief in a child over one — and it stops at the edge of a cold. Regulators assessed the shelf’s larger claims and authorised none, mostly because honey varies too much for any effect to be pinned to it. Two safety lines — no honey under 12 months, and rare but real allergy — are firm. What the evidence leaves honey is what it always had: flavour, variety and history.