Herbal and nutraceutical options

Why there is no single "herbal" answer

These products were grouped together only because researchers compared botanicals studied for anxiety in one analysis. That analysis did not find a shared class effect. It found a top signal, an uncertain middle, and a weak tail, and it described itself as preliminary. The agents differ in extract, dose, mechanism, and risk, so evidence for one does not carry over to another. The only fair way to read the shelf is one divider at a time.

Silexan (standardized lavender oil): the lead signal

Within this network, Silexan (a standardized lavender oil preparation) had the most robust signal. That is a comparison inside a preliminary analysis, held at moderate-to-low confidence. "Most robust here" is not the same as "proven." It leads partly because the agents around it are weaker, not because a large, independent, diagnosed-disorder literature has confirmed it.

Kava: two separate lines that must not merge

Kava is the controversy branch, and it needs two sentences kept apart.

On benefit: kava showed an overall anti-anxiety signal, and a Cochrane review found a modest but significant reduction in anxiety across 12 trials. But that signal is an overall one and is uncertain specifically for generalized anxiety disorder, which is often what people mean by "anxiety."

On safety: the side effects seen in those short trials describe in-trial tolerability only. They do not settle the serious long-term concern about liver toxicity, which later led to regulatory restrictions in several countries. "Kava works and only had mild side effects" wrongly fuses an uncertain benefit claim with an unearned safety claim. The safety question is not cancelled by the benefit signal.

Smaller signals: ginkgo and ashwagandha

Ginkgo biloba and ashwagandha (Withania) each showed a signal in the network, but on small samples. A signal on small samples is a research lead, not an established treatment for a diagnosed anxiety disorder.

Ashwagandha is popular, so it is worth being clear: popularity is not evidence. The stronger evidence that exists for ashwagandha is mostly about short-term perceived stress and anxiety scores, not diagnosed anxiety disorders, and it attaches to specific standardized root extracts at specific doses, not to whatever is on a label or in a multi-herb "stress blend." Case reports of liver injury also exist. For those reasons, ashwagandha stays a named branch here rather than a separate recommendation. You can read the fuller comparison on its dedicated profile.

See the Traditional Therapies ashwagandha profile →

The weaker tail

Passionflower, saffron, valerian, chamomile, L-theanine, and other agents in the network rest on limited, mixed, high-risk, or single-study evidence. None earns a stand-alone treatment claim for a diagnosed anxiety disorder, and none makes any other agent more credible. This tail is not a diluted version of one positive result; it is a set of unresolved, agent-specific questions.

Safety needs dividers too

One reassuring "herbal so it is gentle" sentence cannot cover this shelf.

• Kava carries an unresolved liver-toxicity signal and a history of regulatory restriction.
• Ashwagandha carries case reports of liver injury and wide product variability.
• Sedating botanicals, and those that interact with medicines, need a review with a clinician or pharmacist before use.
• Extract, dose, and preparation are not interchangeable across products, or even across the studies that tested them.

Small, short trials with few side effects do not prove long-term or cross-product safety.

What the evidence does not settle

• Long-term safety and durability for every agent.
• Whether the product on a shelf matches the extract and dose that was studied.
• Whether score changes in stress studies translate to treating a diagnosed disorder.
• How stable a preliminary network ranking will prove to be.