Linalool, And Why A Number Cannot Carry A Claim

Linalool is a monoterpene alcohol and a common constituent of aromatic plants. Published analyses place it at roughly 60 to 80% of coriander seed oil, 20 to 50% of lavender oil and 30 to 50% of sweet basil oil, with wide variation by cultivar, growing region and distillation method. Lavender grown in the United States has been reported in the range of 20 to 43%.

As with any constituent, a percentage quoted without its cultivar and its method is an illustration rather than a specification.

The preparation used in the well-known trials is defined material: an oil of Lavandula angustifolia obtained by steam distillation from fresh flowering tops, given by mouth in soft gelatine capsules at two dose levels, the higher twice the lower. Linalool and linalyl acetate together account for about 70% of its constituents.

The largest trial randomised 539 adults who met DSM-5 diagnostic criteria for the disorder under study, assigning them to the higher or the lower dose of the preparation, an active comparator (a prescription medicine at its usual daily dose), or placebo, once daily for ten weeks, double-blind and double-dummy. Reported decreases on the trial's primary clinician-rated scale were 14.1 points (SD 9.3) for the higher dose, 12.8 (SD 8.7) for the lower, 11.3 (SD 8.0) for the active comparator and 9.5 (SD 9.0) for placebo, with both doses of the preparation separating from placebo at p < 0.01.

A network meta-analysis published in 2019 pooled five randomised controlled trials covering 1,320 participants.

The evidence is real. What it licenses is narrower than it first appears, and it fails to reach a profile figure at three separate points.

The first gap is the substance. No arm tested isolated linalool. Roughly 30% of the material was something other than linalool and linalyl acetate, and no trial has separated the contributions. Attributing the result to one named constituent is an assumption the design cannot carry.

The second gap is the route. The trials used oral capsules. A profile describes material that is not used that way, and nothing in the trials speaks to any other route.

The third gap is the quantity. The trials administered a defined dose of a standardised preparation. A percentage on a profile is a proportion of a sample, not a dose of anything, and the two are not convertible.

Any one of those gaps would be enough on its own.

The gaps above are about study design. Underneath them sits a pharmacological question that is genuinely open, and it deserves stating properly rather than gesturing at.

For linalool specifically to account for those results, it would need to reach the relevant tissue at a concentration where it does something. Candidate mechanisms exist and are not speculative — modulation at GABA-A receptors among them — and the assays behind them are real measurements. The unresolved step is the one that decides the fate of most in-vitro pharmacology: whether the concentration that produces an effect in a dish is ever approached in a person at a realistic exposure. That is a pharmacokinetic question rather than a pharmacological one, and unlike much else here it is answerable in principle with existing methods.

There is a further wrinkle particular to this preparation. Linalyl acetate is the other major constituent, and it is the acetate ester of linalool. Whether it acts in its own right, or functions as a reservoir hydrolysed to linalool after absorption, or does both, is precisely the sort of thing a trial comparing whole preparation against placebo cannot separate. The trial was not built to, and it is no criticism of the trial that it did not.

Those are the questions worth putting to this literature. They are considerably more specific than "more research is needed", and they are the ones whose answers would actually move the position.

This is not a quirk of one compound. It is the standard shape of the evidence across this entire field.

Where good human trials exist for aromatic plant constituents, they have almost always tested standardised whole preparations, because that is what can be manufactured consistently and blinded reliably. Where isolated compounds have been studied, it has mostly been in cell culture or in animals, at exposures and by routes that do not correspond to human use.

So the evidence base is shaped like mixtures, and a profile is shaped like a list of isolated constituents. A claim cannot cross from one shape to the other without an assumption nobody has tested. That is the whole reason a profile is a composition measurement rather than a prediction.

The pillar article makes the general point; linalool is a clear instance of it.

The compound autoxidises on contact with air. Skold and colleagues identified a linalool hydroperoxide as the reactive species in 2002, and in a ten-year consecutive patch-test series of 6,719 patients published in 2026, hydroperoxides of linalool produced positive reactions in 3.1% of patients, against 1.6% for hydroperoxides of limonene. European cosmetics regulation requires linalool to be declared as a fragrance allergen above 0.001% in leave-on and 0.01% in rinse-off products.

The point here is not dermatological. It is that a figure reported at testing describes a compound that is still reacting, and the material in front of you is further from that figure the longer it has been sitting.

A randomised comparison of isolated linalool against the standardised whole preparation, at matched exposure, on the same instruments, would establish what the single constituent contributes. It has not been run.

Until it is, the precise statement is this: a standardised whole preparation has repeatedly separated from placebo on the rating scales those trials used, and what linalool specifically contributed to that result is unmeasured.

Those are two different sentences, and a figure on a profile is a third thing again. Most writing on this subject prints the first while naming the subject of the second. Telling the three apart is not pedantry — it is the difference between a supplier's marketing and an assessment worth relying on.