Do terpenes predict effects?

Where the idea comes from

The belief that terpenes steer effects mostly traces back to the “entourage effect”: the hypothesis that the compounds in cannabis, terpenes included, modify one another’s activity. An influential 2011 review set it out as a set of possibilities worth testing (Russo, 2011). Since then it has become a marketing staple, and menus and packaging routinely pair individual terpenes with promises about how a product will make you feel.

The hypothesis is legitimate science. The promises run far ahead of it.

What the research has tested

The evidence comes in three kinds, and it doesn’t agree with itself.

• Cells. Two independent laboratory studies tested common cannabis terpenes, including myrcene, limonene, α- and β-pinene and β-caryophyllene, on cells carrying the human cannabinoid receptors CB1 and CB2. Neither found evidence that the terpenes act at those receptors or change how cannabinoids act there (Santiago et al., 2019; Finlay et al., 2020). A third group reported the opposite in its own cell experiments (LaVigne et al., 2021).
• Animals. That same group found that some terpenes produced cannabinoid-like behaviors in mice and added to a synthetic cannabinoid’s activity, at injected doses of 50 to 200 milligrams per kilogram of body weight (LaVigne et al., 2021). That’s a reasonable way to probe biology. It isn’t a model of smelling or inhaling flower.
• People. Controlled human studies are only beginning. In one, 20 healthy adults inhaled a purified terpene, d-limonene, from a vaporizer alongside THC, at doses the researchers set and under double-blind conditions (Spindle et al., 2024). It’s a careful study, and it’s one small study. It used a purified compound at a known dose, which is exactly what a jar and its COA can’t give you.

Why a lab report can’t be turned into a prediction

Even if the biology were settled, a COA wouldn’t get you there.

• A percentage isn’t a dose. The report tells you what share of a tested sample was, say, limonene. It doesn’t tell you how much of anything reaches you, which depends on how much you use, how you use it and what heat does to the compounds on the way.
• The panel is partial, and the jar has changed. Standard panels measure a list of terpenes, not every compound, and levels keep falling after the test date (see how to read a COA).
• Most of what shapes an experience isn’t on the panel. How much THC is taken and how, a person’s tolerance and history, their expectations and the setting all vary from person to person and session to session. None of them appears on a terpene panel.
• The labels don’t help. The indica and sativa labels most often used to suggest effects don’t match measured chemistry or genetics (Smith et al., 2022; Watts et al., 2021).

What NOSE does instead

NOSE compares one thing: how closely the measured terpene proportions of two jars line up. A high score means their profiles are similar in shape, which is the part of their aroma a lab panel can see. It doesn’t mean they’ll feel alike, and we don’t describe any jar in terms of how it will make you feel.

Even the aroma claim is held loosely. The idea that a similar terpene shape predicts you’ll enjoy a jar is a hypothesis we’re testing, not a result, which is why every match has a thumbs up and a thumbs down.

If you use cannabis for a health reason, talk to a clinician. A terpene panel is not medical advice, and neither is NOSE.

What would change our mind

If well-designed human studies show that terpene levels at the amounts found in flower reliably predict what people experience, we’ll say so here and cite them. Until then, we’d rather under-promise.