Lemon Balm: Independent Evidence on Anxiety, Sleep & Safety

Table of contents

• Evidence summary
• What lemon balm is
• How it works
• What works and what does not
• Benefits by claim
• Combination products and attribution limits
• Risks and all side effects
• All interactions
• Who should avoid lemon balm
• Doses studied and product quality
• Follow the money: who makes it and who funded the evidence
• Related research
• Frequently asked questions
• Sources and funding notes

Evidence summary

Independent evidence and credibility scorecard

Independence tiers: 1 = regulator or government body; 2 = academic or Cochrane team with no declared commercial funding for the work; 3 = independent authors analysing mostly manufacturer-run trials, or minor declared industry ties; 4 = manufacturer-funded or manufacturer-authored. Credibility runs from A (highest) to D.

Sources cluster in two places: small academic trials from Iran, and industry-linked research from Western Europe (France, Switzerland, Italy, Germany). We found no large, publicly funded trial of lemon balm for anxiety or sleep in the UK, US or Canada.

What lemon balm is

Lemon balm (Melissa officinalis, also called balm, common balm or melissa) is a lemon-scented perennial herb. Branded extracts are usually standardised to rosmarinic acid. Cyracos, for example, is standardised to more than 7% rosmarinic acid (Cases et al. 2011).

Forms on sale

• Tea and dried leaf: the traditional form covered by the EU herbal monograph.
• Branded standardised extracts: Cyracos (Naturex, now Givaudan), Zensera (Givaudan), Relissa Phytosome (Indena) and Bluenesse (produced by Cognis and sold by Vital Solutions). These differ in solvent, strength and carrier, so a result from one cannot be assumed for another.
• Combination products: often with valerian, passionflower, hops or butterbur.
• Cream: a 70:1 extract cream (Lo-701) was tested for cold sores (Koytchev et al. 1999).

Legal status

• EU: the HMPC monograph (adopted 14 May 2013, first published 5 August 2013) lists lemon balm as a traditional herbal medicine for "relief of mild symptoms of stress and to aid sleep", based on long-standing use (EMA). The EMA lists the monograph as under periodic review, with a call for scientific data in 2024 (EMA).
• Canada: secondary sources report that Health Canada recognises lemon balm as a traditional herbal medicine for restlessness due to mental stress and sleep. We did not access the monograph text.
• US: sold as a dietary supplement, which does not require pre-market proof of safety or efficacy. It is not FDA-approved to treat any condition.

How it works

No mechanism has been proven in people. The proposed mechanisms come from cell, tissue and animal work:

• GABA breakdown: Cyracos was described as containing phytochemicals that inhibit GABA catabolism (Cases et al. 2011). Whether this happens in people is unknown.
• Cholinergic binding: leaf preparations differ in how strongly they bind nicotinic and muscarinic receptors in human brain tissue. The extract in one earlier study showed no such binding, so results cannot be carried over between preparations (Kennedy et al. 2003).

What is shown in people is limited to symptom changes in trials. Which compound causes them, and whether a cup of tea contains enough, is unknown.

What works and what does not

Benefits by claim

Stress and "calm" after a single dose

Northumbria University's cognition group ran the best-known single-dose studies. In 20 young adults, a Pharmaton extract at 300 mg raised self-rated calmness at the lowest dose, while 900 mg reduced alertness at every time point, and 600 mg improved sustained accuracy of attention (Kennedy et al. 2002). In 18 volunteers put through a laboratory stress test, 600 mg improved calmness and reduced the negative mood the test caused, while 300 mg sped up mental arithmetic (Kennedy et al. 2004). A later study tested single doses of 600, 1,000 and 1,600 mg of a dried-leaf sample in 20 volunteers (Kennedy et al. 2003).

The EMA page notes that the clinical studies had design shortcomings, so the traditional-use status rests on long-standing use rather than proven efficacy (EMA).

A 2026 acute trial of Givaudan's Zensera extract (300 mg, single dose) enrolled 130 young adults, and 106 completed it. Its primary outcome, overall calmness, did not change. Secondary analyses found better performance on the hardest attention tasks at 5 hours (Mathews et al. 2026). The study received funding from Givaudan France Naturals as a PhD studentship, and two co-authors were Givaudan employees. Food-based versions (a beverage and a yoghurt) were tested in healthy young adults in a study sponsored by Rudolf Wild and Cognis (Scholey et al. 2014).

Anxiety

A meta-analysis of randomised trials searched 12 databases to October 2020 and found that lemon balm improved anxiety scores against placebo, SMD −0.98 (95% CI −1.63 to −0.33). The authors said the results "should be interpreted with caution" because of high heterogeneity and few trials (Ghazizadeh et al. 2021). We could not access the full text to check the trial list.

Individual trials are small. In 80 patients with stable angina, 3 g a day for 8 weeks produced significant reductions in depression scores compared with placebo (Haybar et al. 2018). The first Cyracos study reported an 18% fall in anxiety scores in 20 stressed volunteers, but it was open-label and had no placebo group (Cases et al. 2011).

A 3-week trial in 104 randomised adults with emotional distress (100 completed) gave 400 mg a day of Indena's Relissa phospholipid extract. Depression scores fell from 23.3 to 6.9 on the supplement and from 22.7 to 17.1 on placebo (Bano et al. 2023). Effects this large over three weeks warrant caution. Indena supplied the product and placebo tablets, and the authors declared no conflicts of interest.

We found no trial of lemon balm in people formally diagnosed with generalised anxiety disorder that compared it with a standard treatment such as an SSRI or psychological therapy.

Depression and low mood

The same meta-analysis found a moderate effect on depressive symptoms, SMD −0.47 (95% CI −0.73 to −0.21), in people with symptoms of depression or anxiety rather than diagnosed major depression (Ghazizadeh et al. 2021). In a 12-week trial of 60 people with type 2 diabetes and depressive symptoms, 700 mg a day of a hydroalcoholic extract improved depression and anxiety scores compared with a toasted-flour control. Only 44 completed the trial (Safari et al. 2023).

Sleep and insomnia

Sleep is the main traditional use, but trials are small. A preprint systematic review posted in March 2026 reports a lemon balm subgroup effect (SMD −0.66). It states that most randomised trials had high risk of bias, and its abstract gives no confidence interval or trial count for the lemon balm subgroup. The preprint is not peer reviewed; it reports no specific grant and no competing interests (Paracha et al. 2026).

In the Bano trial, sleep scores (PSQI, higher is worse) fell from 10.9 to 3.5 on lemon balm and from 8.6 to 6.6 on placebo. The lemon balm group started with significantly worse sleep, which favours a larger fall (Bano et al. 2023). In the 12-week type 2 diabetes trial, sleep quality did not differ significantly between groups (Safari et al. 2023).

A University of Reading sleep trial with Naturex as a collaborator (NCT05422599) enrolled 99 people and completed in May 2023. Its registry record shows no posted results.

Heart palpitations

In Persian traditional medicine lemon balm is used for palpitations. In a double-blind trial, 500 mg twice a day of a freeze-dried water extract for 14 days reduced palpitation episodes and the number of anxious patients compared with placebo. 55 of 71 recruited adults completed it (Alijaniha et al. 2015). Palpitations can signal heart-rhythm problems; see our guide to palpitation causes and assessment.

Premenstrual symptoms

One double-blind trial in 100 Iranian high-school girls gave 1,200 mg a day over three cycles and reported lower premenstrual symptom scores than placebo (Akbarzadeh et al. 2015). These are single-centre findings without replication.

Menopause

A trial in 100 women aged 50 to 60 with sleep disorders compared a valerian/lemon balm combination with placebo and reported less sleep disturbance in the combination group (Taavoni et al. 2013). Because the product combined two herbs, it cannot show lemon balm's effect alone.

Memory, focus and dementia

In a 96-week trial of an extract providing 500 mg of rosmarinic acid daily, 323 older adults with subjective or mild cognitive impairment showed no significant difference from placebo on the primary cognitive score. In a subgroup without hypertension, a difference favoured the extract (p = 0.036). The authors describe this as a possible benefit that needs confirmation (Noguchi-Shinohara et al. 2023).

Cold sores: a different use

Topical lemon balm is a separate product. A 70:1 extract cream (Lo-701) was tested in a 66-patient trial. On day 2 its combined symptom score was lower than placebo (Koytchev et al. 1999). This says nothing about oral use for mood or sleep.

Combination products and attribution limits

Many products combine lemon balm with other sedative herbs. A positive trial of a mix is evidence for that mix, not for lemon balm alone.

• Ze 185 (lemon balm, valerian, passionflower and butterbur): in 182 patients with somatoform disorders, the four-herb product and a three-herb version without butterbur both beat placebo over two weeks, with the four-herb product best (Melzer et al. 2009). Because the trial tested combinations, it cannot isolate lemon balm's share.

See our separate reviews of valerian and chamomile for the companion herbs.

Risks and all side effects

Short-term lemon balm appears well tolerated in trials. Headache, dizziness and bloating are the minor effects reported, often at similar rates on placebo (NIH LiverTox). Drowsiness and reduced alertness are the main concern: they were seen at higher single doses in healthy volunteers (Kennedy et al. 2002). The EMA page notes that no side effects had been reported with these medicines when it assessed them (EMA). We could not access the monograph's full list of undesirable effects.

Thyroid: a lab signal with unknown clinical meaning

In laboratory work, freeze-dried water extracts of lemon balm inhibited thyroid-stimulating hormone (TSH) binding to human thyroid membranes and its effect on adenylate cyclase. The inhibition did not persist after the membranes were washed, which suggests an interaction with the hormone rather than the receptor (Auf'mkolk et al. 1984). A 1985 study reported similar inhibition of thyroid-stimulating antibody activity (Auf'mkolk et al. 1985), and an Italian assay confirmed inhibition of TSH binding to its receptor (Santini et al. 2003). The EU monograph notes these findings and states that their clinical relevance is not known (EMA monograph). We found no controlled human study showing that lemon balm changes thyroid blood tests. People with thyroid disease, or on thyroid medicines, should tell their clinician before using it.

All interactions

We found no controlled interaction studies for lemon balm. The table lists theoretical overlaps, based on how the herb acts and on the sedation shared by herbal sedatives. None has been tested in people.

Tinctures and liquid extracts often contain alcohol, which should be declared on the label.

Who should avoid lemon balm

• Pregnancy and breastfeeding: we found no adequate human data on safety in pregnancy or while breastfeeding. Avoid unless a clinician advises otherwise.
• Children under 12: the EMA page says the herb should only be used in adults and children over 12 (EMA).
• Thyroid disease or thyroid medicines: caution because of the lab TSH signal; discuss with your clinician.
• Before surgery or anaesthesia: because of possible additive drowsiness, tell the surgical team about any herbal sedative.
• Drivers and machine operators: do not drive if it makes you drowsy.
• Known allergy to lemon balm.
• Liver or kidney disease: we found no specific data. LiverTox reports no liver-injury signal, but we found no trials in people with liver or kidney disease.
• People with significant anxiety, depression or insomnia: lemon balm is not a substitute for assessment. Treatments with stronger evidence include CBT-I.

Doses studied and product quality

These are trial doses, reported for context. They are not a recommendation.

Research doses vary widely because extracts differ in solvent and concentration.

Product quality and contamination

• Pyrrolizidine alkaloids: these liver-toxic plant chemicals can enter herbal teas through weeds harvested with the crop. Germany's BfR tested 221 herbal tea and tea samples in 2013. Total pyrrolizidine alkaloid content was up to 3.4 mg/kg across the samples, a figure that is not specific to lemon balm, and the BfR called the levels "too high" (BfR 2013). EU Regulation (EU) 2020/2040 set a 400 µg/kg limit for dried lemon balm infusions from 1 July 2022 (EUR-Lex). That regulation has since been replaced by Regulation (EU) 2023/915, and we did not re-check the current limit.
• Label accuracy: US supplements do not require pre-market proof of safety or efficacy, and labelled and actual contents can differ.
• Extracts are not interchangeable: the extract in one earlier study showed no cholinergic binding, while other preparations did, so results from one preparation may not apply to another (Kennedy et al. 2003).

Follow the money: who makes it and who funded the evidence

Who makes and sells it

• Givaudan (Vernier, Switzerland): completed its takeover of Naturex (Avignon, France) on 18 September 2018, after a squeeze-out procedure, having held 98.06% of the shares (Givaudan 2018). Naturex made Cyracos (Cases et al. 2011). Givaudan France Naturals made the Zensera extract used in the 2026 acute trial (Mathews et al. 2026).
• Indena S.p.A. (Milan, Italy): makes Relissa, and supplied the product and placebo for the Bano trial (Bano et al. 2023).
• Cognis and Rudolf Wild (Germany): sponsored the food-based studies. Cognis produced the standardised extract, which is sold as Bluenesse (Scholey et al. 2014).

Who funded the trials

• Extract makers: several authors of the first Cyracos study had Naturex affiliations. The paper acknowledges financial assistance from Naturex and declares no conflict of interest (Cases et al. 2011). It was open-label with no placebo group.
• Zensera: the 2026 trial received funding from Givaudan France Naturals as a PhD studentship, and two co-authors were Givaudan employees. The paper states that the funder had no role in design, analysis or publication (Mathews et al. 2026).
• Supplied product: Indena supplied Relissa and placebo tablets for the Bano trial, whose authors declared no conflicts (Bano et al. 2023).
• Sponsored food studies: Rudolf Wild and Cognis sponsored the Scholey 2014 studies. The sponsors contributed to study design and outcome choice (Scholey et al. 2014).
• Public money: an NIH RePORTER search for "lemon balm" returned zero projects (checked 9 October 2026; NIH RePORTER).
• Regulators and reference resources: NIH's LiverTox is a US government resource (LiverTox). The EMA is funded mostly by fees paid by industry: about 91.5% of its 2026 budget of €615.5 million (EMA). Its herbal committee judged the clinical evidence insufficient to establish well-established use.

Documented integrity events

• Registered trial without published results: the University of Reading sleep trial (NCT05422599) lists Naturex as a collaborator, enrolled 99 people and completed on 23 May 2023. Its registry record shows no posted results. Unpublished trials can bias the evidence if negative results stay unseen. We do not know why it is unpublished.
• FDA warning letter: in November 2020 the FDA warned Red Moon Herbs over products marketed for COVID-19, including one that combined St John's wort, lemon balm and calendula (FDA).
• Retraction: a 2023 review of lemon balm's antiviral potential was retracted on 30 October 2025 for textual overlap with earlier papers without sufficient attribution. The authors did not agree with the decision (retraction notice).

What this means: regulators accept lemon balm only on the strength of tradition. The independent trials are mostly small Iranian studies, and the branded-extract trials are mostly funded, authored or supplied by the companies that sell the extracts. Both groups point in the same direction (modest calming and sleep benefits), which is mildly reassuring, but neither is strong enough to say how much lemon balm helps or for whom.

Related research

• Stress, anxiety and depression supplements: evidence and safety
• Generalised anxiety disorder: treatments, supplement evidence and safety
• Insomnia: assessment, treatments, supplement evidence and safety
• Sleep supplements: evidence limits, product quality and safety
• Valerian: sleep evidence, safety and funding limits
• Chamomile for sleep and anxiety
• Oral lavender and Silexan
• Ashwagandha: evidence for stress, sleep and safety
• L-theanine: calm focus and anxiety evidence
• Premenstrual dysphoric disorder (PMDD)
• Heart palpitations: causes and assessment
• CBT-I: cognitive behavioural therapy for insomnia

Frequently asked questions

Sources and funding notes

Key sources, with what we checked against the original record:

1. EMA herbal medicine page for Melissae folium (traditional-use wording, periodic review status, use only in adults and children over 12) and the EMA/HMPC monograph (2013). Regulator; EU.
2. Ghazizadeh J et al. Lemon balm on depression and anxiety. Phytother Res 2021. Authors at Iranian and Australian institutions; competing interests declared as none; funding statement not seen.
3. Paracha MA et al. Lesser known herbal sedatives for insomnia. medRxiv 2026 (preprint, not peer reviewed). No specific grant; no competing interests.
4. Kennedy DO et al. 2002, 2003 and 2004. Northumbria University, UK. Funding statements not checked.
5. Cases J et al. Pilot trial of Cyracos. Med J Nutrition Metab 2011. Open-label, 20 volunteers; Naturex-affiliated authors; financial assistance from Naturex acknowledged; conflict of interest declared as none.
6. Mathews I et al. Zensera acute trial. Ther Adv Psychopharmacol 2026. Funded by Givaudan France Naturals (PhD studentship); two Givaudan employees; 130 enrolled, 106 completed; registered as NCT06183372.
7. Bano A et al. Relissa trial. Front Pharmacol 2023. 104 randomised, 100 completed; product and placebo supplied by Indena S.p.A.; no conflicts declared.
8. Scholey A et al. Anti-stress effects of lemon balm-containing foods. Nutrients 2014. Sponsored by Rudolf Wild GmbH and Cognis GmbH.
9. Melzer J et al. Ze 185 in somatoform disorders. Phytother Res 2009. 182 patients.
10. Alijaniha F et al. Heart palpitations. J Ethnopharmacol 2015.
11. Akbarzadeh M et al. Premenstrual syndrome. Nurs Midwifery Stud 2015.
12. Taavoni S et al. Valerian/lemon balm in menopause. Complement Ther Clin Pract 2013.
13. Haybar H et al. Chronic stable angina. Clin Nutr ESPEN 2018.
14. Safari M et al. Type 2 diabetes with depression. BMC Complement Med Ther 2023. 60 randomised, 44 completed.
15. Noguchi-Shinohara M et al. Older adults without dementia. J Alzheimers Dis 2023.
16. Koytchev R et al. Lo-701 cream for herpes labialis. Phytomedicine 1999. 66 patients.
17. Auf'mkolk M et al. 1984; Auf'mkolk M et al. 1985; Santini F et al. J Endocrinol Invest 2003.
18. NIH LiverTox: Lemon Balm (updated 10 February 2024). Likelihood score E.
19. ClinicalTrials.gov NCT05422599. University of Reading, with Naturex as collaborator; enrolled 99; completed 23 May 2023; no results posted.
20. Givaudan: completion of the Naturex acquisition and delisting (18 September 2018; 98.06% held before the squeeze-out).
21. FDA warning letter to Red Moon Herbs (18 November 2020).
22. Retraction notice, Nutr Metab Insights (30 October 2025), for Behzadi et al. 2023.
23. BfR press release on pyrrolizidine alkaloids in herbal teas (2013).
24. Commission Regulation (EU) 2020/2040 (applicable from 1 July 2022; since replaced by Regulation (EU) 2023/915).
25. NIH RePORTER search for "lemon balm": zero projects (checked 9 October 2026).

What we could not verify

The full text and funding statement of the Ghazizadeh meta-analysis, so we do not state its trial count. The funding statements of most Northumbria, Iranian and Japanese trials. The EMA monograph's wording on pregnancy, interactions, driving and undesirable effects, which we could not access in full. The LactMed and Health Canada monograph texts, which we did not access directly. The current EU limit for pyrrolizidine alkaloids in lemon balm (Regulation 2023/915). The confidence interval and trial count for the lemon balm subgroup in the Paracha preprint. The results of NCT05422599 beyond its registry status. A reliable market-size figure for lemon balm.