Vanilla extract: at least 35% alcohol, and a small allergy signal
The human studies on vanilla are about allergy, and they are small. In 11 children under five with severe eczema, blinded oral challenges with balsam of Peru, 50 mg of natural vanilla and 12.5 mg of vanillin produced eczema reactions in 9. The 1994 study is tiny, and its challenge mix does not separate vanilla from balsam of Peru. In 22 adults with delayed allergy to balsam of Peru, a blinded challenge with spices including vanillin sugar raised the number of palm blisters by 30 to 280% in 8 people, with no such rise on placebo. Vanilla extract itself was not tested alone. For everyone else the fact to know is the alcohol. Under US rules, a product sold as vanilla extract must contain at least 35% ethyl alcohol by volume, while imitation vanilla and vanilla flavoring fall under other sections.
Every statement here is tied to a source. The badge says what kind of evidence stands behind it: A is a meta-analysis or systematic review, E is the position of an expert body without its own analysis. How we verify →
The numbers, per 3.5 oz (100 g)
| USDA entry | kcal | Protein | Fat | Carbs | Fiber |
|---|---|---|---|---|---|
| Vanilla extract, imitation, no alcohol | 56 | 0.03 g | 0 g | 14.4 g | 0 g |
| Vanilla extract | 288 | 0.06 g | 0.06 g | 12.7 g | 0 g |
| Vanilla extract, imitation, alcohol | 237 | 0.05 g | 0 g | 2.41 g | 0 g |
These are the USDA entries that are vanilla extract, not foods made from it; they differ in cut, part, pack or preparation. Linked names have a page with the full profile and per-portion figures.
How much is actually in it
- Lab extraction measured about 47-50 mg of vanillin per gram of vanilla beans and used coumarin as the internal standard because vanilla beans contain none. — vanilla beans and tonka beans, laboratory Evidence D sourceVanillin 47.4 mg/g (subcritical water, 150 C, 8 h) vs 49.7 mg/g (sonication)
- Under US rules, vanilla extract must contain at least 35% ethyl alcohol by volume and at least one unit of vanilla constituent per gallon. — products sold as vanilla extract in the US Evidence E sourceEthanol at least 35% by volume
What cooking and storage change
- In lab heating tests, 93-98% of vanillin was recovered after 60 minutes in hot pressurized water at 100-200 C. — pure vanillin in water, laboratory Evidence D sourceVanillin recovery after 60 min: 93% at 100 C, 98% at 150 and 200 C
Safety, limits and interactions
- In 22 adults with delayed allergy to balsam of Peru, blinded oral challenges with spices including vanillin sugar raised palm blisters by 30-280% in 8, with no such rise on placebo. — 22 patients with delayed-type allergy to balsam of Peru (n = 22) Evidence B source8 of 22 had a 30-280% rise in palmar vesicles after active substances, none of them after placebo
- The FAO/WHO expert committee JECFA kept an acceptable daily intake of 0-10 mg/kg body weight for vanillin in 2001 and saw no safety concern at current flavoring use in 2019. — general population, vanillin as flavoring Evidence E sourceADI 0-10 mg/kg bw
What people believe that the data does not support
- US law treats any food with added coumarin, including tonka bean or tonka extract, as adulterated, a rule in force since 1954. — all human food sold in the US Evidence E sourceAdded coumarin prohibited
Who this works differently for
- In 11 children under five with severe eczema, blinded oral challenges with balsam of Peru, 50 mg natural vanilla and 12.5 mg vanillin caused eczema in 9. — 11 children under five years with severe atopic dermatitis (n = 11) Evidence C source9 of 11 reacted with eczema, 1 with Quincke edema, 2 negative, 6 improved after avoiding flavorings
- In about 150 premature infants, vanilla odor before feeds came with 11.8 days to full oral feeding, against 12.2 days with rose odor and 12.9 days with no odor. — about 150 premature infants over 27 weeks gestation on tube feeding (n = 150) Evidence B source11.8 (vanilla) vs 12.2 (rose) vs 12.9 (control) days to full oral feeding
- A systematic review found moderate evidence that infants can detect vanilla and other flavors in breast milk within hours of a single maternal intake. — studies of lactating mothers and infants, 1980 to 2017 (n = 15 articles on lactation) Evidence C sourceFlavor detected within hours of one maternal ingestion (moderate evidence grade)
- A 42-year-old man with alcohol dependence drank 20 fl oz (600 mL) of vanilla essence and developed status epilepticus, severe metabolic acidosis and a hyperosmolar coma. — one 42-year-old man with ethanol dependence (n = 1) Evidence D sourceStatus epilepticus, severe metabolic acidosis, osmolar gap 151, full recovery after dialysis
The bottom line
The authors of one case report note that people with alcohol dependence may drink vanilla extract for its alcohol. In that report a 42-year-old man with alcohol dependence drank 20 fl oz (600 mL) of vanilla essence and developed prolonged seizures, severe acidosis and a coma, and recovered after dialysis. It is a single case. Vanillin, the main flavor compound, has an acceptable daily intake of 0 to 10 mg per kg of body weight, set in 1967 and kept by the FAO/WHO expert committee JECFA in 2001. JECFA sees no safety concern at current intakes as a flavoring. That judgement covers vanillin and says nothing about the alcohol in the bottle. Coumarin is a separate question. US law has treated food with added coumarin, including coumarin from tonka beans or tonka extract, as adulterated since an order of 1954. Chemists measuring one sample of vanilla beans used coumarin as a marker because vanilla beans contain none, and found about 47 to 50 mg of vanillin per gram of bean. Heat does little to vanillin in the lab. After 60 minutes in hot pressurized water, 93% of pure vanillin was recovered at 100 C and 98% at 150 and 200 C. That was water, not a cake batter, and the study did not measure the alcohol. Who should be careful. Children with severe eczema and adults allergic to balsam of Peru are the groups the challenge studies point to. Anyone avoiding alcohol should count vanilla extract as a product that is at least 35% alcohol. A systematic review found moderate evidence that infants can detect vanilla and other flavors in breast milk within hours of one maternal intake, a qualitative synthesis that reports detection and no harm or benefit. In about 150 tube-fed premature infants, smelling vanilla before feeds came with 11.8 days to full oral feeding, against 12.2 with rose odor and 12.9 with none, a small difference from odor alone with no extract swallowed.
Related foods
More from the same food group (spices and herbs)
- Vinegar and blood sugar: one narrow effect
- Allspice: a pinch of flavor with no trial of its own
- Anise seed: one striking trial, never repeated
- Turmeric: small capsule effects, and a spice nobody has tested
- Thyme: big numbers per 3.5 oz (100 g), almost no trials of the herb itself
- Spearmint: capsule results, a tea trial and an oil allergy
Sources
- vex-r5-07 · lab analysis
Since there is no coumarin in vanilla beans, it was used as the internal standard for separation and determination of vanillin. ... The vanillin quantity obtained by SBWE under this condition is 47.4 mg/g, which is very close to the 49.7 mg/g achieved by sonication extraction.
Molecules, 2020 · checked 2026-09-29 - vex-r5-08 · regulation
In vanilla extract the content of ethyl alcohol is not less than 35 percent by volume and the content of vanilla constituent, as defined in § 169.3(c), is not less than one unit per gallon.
21 CFR 169.175, Vanilla extract (eCFR; text read from govinfo CFR 2024 edition) · checked 2026-09-29 - vex-r5-06 · lab stability study
The recovery of vanillin after heating for 60 min ranges from 93% at 100 °C to 98% at both 150 and 200 °C, as shown in Table 1. Thus, no significant degradation of vanillin occurred under the subcritical water conditions investigated.
Molecules, 2020 · checked 2026-09-29 - vex-r5-02 · double-blind challenge
Double-blind placebo-controlled peroral challenges (DBPCPCs) with balsam of Peru and spices (active substances) were performed on 22 patients. A substantial increase (30-280%) in the number of palmar vesicles after the DBPCPC with the active substances, but not with placebo, was seen in 8 patients
Contact Dermatitis, 1995 · checked 2026-09-29 - vex-r5-10 · agency evaluation
No safety concern at current levels of intake when used as a flavouring agent. The 1967 ADI of 0-10 mg/kg bw was maintained at the fifty-seventh meeting (2001).
WHO JECFA database, Vanillin · checked 2026-09-29 - vex-r5-09 · regulation
Food containing any added coumarin as such or as a constituent of tonka beans or tonka extract is deemed to be adulterated under the act, based upon an order published in the Federal Register of March 5, 1954 (19 FR 1239).
21 CFR 189.130, Coumarin (eCFR; text read from govinfo CFR 2024 edition) · checked 2026-09-29 - vex-r5-01 · double-blind challenge
Double-blind oral provocation tests were carried out with balsam of Peru (225 mg), natural vanilla (50 mg), artificial vanillin (12.5 mg). Nine children out of eleven presented eczematous reactions, one a Quincke's oedema.
Allerg Immunol (Paris), 1994 · checked 2026-09-29 - vex-r5-03 · RCT
About 150 children born at a postnatal age of 9.5 ± 7.8 days were included in this study. The duration until complete oral nutrition was reached after 11.8 ± 7.7 (vanilla), 12.2 ± 7.7 (rose), and 12.9 ± 8.8 (control) days.
Chem Senses, 2018 · checked 2026-09-29 - vex-r5-04 · systematic review
Moderate evidence indicates that infants can detect diet-transmitted flavors in breast milk within hours of a single maternal ingestion (alcohol, garlic, vanilla, carrot)
Am J Clin Nutr, 2019 · checked 2026-09-29 - vex-r5-05 · case report
Patients with ethanol dependence may consume vanilla extract for the high alcohol content. CASE PRESENTATION: A 42-year-old male with ethanol dependence who developed status epilepticus, severe anion gap metabolic acidosis and a hyperosmolar non-ketotic coma after ingestion of 600 mL vanilla essence.
Clin Toxicol (Phila), 2022 · checked 2026-09-29
Review. Reviewed on 2026-09-29 for evidence level, dose and population context, and claims a reader could misread. Bioma Learn has no human medical reviewer at this time, and we say so rather than invent one. Methodology. This is information, not medical advice.