Ashwagandha Works Modestly. Its LiverTox Score Is B.

We published a piece on lion’s mane recently that ended with a compliment: LiverTox rates it E, the best score in the system, with no case reports of liver injury at all.

Ashwagandha sits on the same shelf, in the same shops, with the same adaptogen framing. Its score is B.

What B means

LiverTox is the NIH’s database of drug-induced liver injury, and its likelihood score rates how confident the evidence is that a substance causes clinically apparent liver injury. B is “likely cause.”

The picture it describes is consistent enough to recognise:

Onset	2 to 12 weeks after starting
Pattern	Usually cholestatic or mixed
Signs	Jaundice and pruritus (itching)
Course	Usually mild to moderate, self-limited
Recovery	Within 1 to 4 months of stopping

And the tail:

Rare instances of fatal liver injury or need for emergent liver transplantation have been reported particularly in patients with preexisting liver disease and cirrhosis.

LiverTox’s practical instruction is to avoid it in patients with cirrhosis or advanced chronic liver disease, and never to rechallenge with the same product.

The part that should change how you read supplement safety claims

Here is LiverTox on what the trials found:

In clinical trials, there have been no reports of serum enzyme elevations occurring during therapy and no mention of serious adverse events or hepatotoxicity.

None. The trials were clean.

The 2021 sleep meta-analysis says the same: no serious side effects reported, all participants completing.

The problem is that five trials and 400 people cannot see a reaction that happens to a small fraction of users. The meta-analysis authors said so themselves, in the conclusion, and it reads differently now than it did then:

Data on the serious adverse effects of Ashwagandha extract are limited, and more safety data would be needed to assess whether it would be safe for long-term use.

They asked for more safety data. It arrived, from the population rather than from a trial, as case reports of jaundice.

This is the general lesson, not an ashwagandha lesson. “No adverse events in clinical trials” is a much weaker statement than it sounds when the trials are small and the reaction is rare. Idiosyncratic liver injury is exactly the kind of thing that only becomes visible after a product goes mainstream.

Was it really the ashwagandha?

A fair question, because contamination is the standard defence for herbal products. LiverTox raises it honestly:

Because commercial herbal preparations are often mixtures of herbs and nutritional products and can be mislabeled and contain unknown herbs and medications, it is not always clear whether the reported cases were due to ashwagandha and one of its components or to a contaminant.

And then closes it:

In several reported cases, however, the commercial product being taken was tested and found to have ashwagandha without other contaminants.

The mechanism is still unknown. The withanolides, the compounds the standardised extracts are standardised for, are the most suspect.

What it does do

The efficacy case is real, and it is modest.

A meta-analysis of five RCTs, 400 participants:

• Overall sleep: SMD −0.59 (95% CI −0.75 to −0.42), I² = 62%
• Larger effects in: diagnosed insomnia, ≥600 mg/day, ≥8 weeks
• Also improved: mental alertness on rising, anxiety level
• No significant effect on quality of life

That last line deserves a moment. Sleep scores moved; how people rated their lives did not. For a product marketed almost entirely on how it will make you feel day to day, an outcome measuring exactly that failing to move is worth knowing.

An SMD of −0.59 is a genuine small-to-moderate effect and better than most of this category manages. It is not nothing. It is also not what the packaging implies.

Reading the two together

This is not a page telling you not to take ashwagandha. It is a page insisting the two facts sit side by side, because they normally do not.

In favour: a replicated, modest, real effect on sleep, best documented at 600 mg or more for eight weeks or more in people with diagnosed insomnia.

Against: a likelihood score of B, a 2 to 12 week window in which cholestatic injury can appear, and a rare tail that has included transplant and death, concentrated in people whose livers were already compromised.

If you take it:

1. Do not take it if you have cirrhosis or advanced chronic liver disease. LiverTox says avoid, not be careful.
2. Know the 2 to 12 week window and what to look for in it: yellowing of the eyes or skin, dark urine, pale stools, persistent itching.
3. Stop at the first sign and see a doctor. Recovery is usual within 1 to 4 months of stopping, and the injury does not appear to become chronic.
4. Never restart the same product after a reaction.
5. Tell your doctor you are taking it, particularly before any blood test that includes liver enzymes.

Why there is no product recommendation on this page

Because the honest summary is “modest benefit, real and occasionally serious liver risk, mechanism unknown,” and that is not a sentence that should end with a button.

We put a purchase box on the lion’s mane article because its LiverTox score is E and its risk file is empty. The difference between those two decisions is the whole point of publishing the score.