Supplement

Bacopa Monnieri

Q: How long does it take for Bacopa to work?

In the studies, the effects are typically measured after twelve weeks of daily intake. The first large controlled study tested after five and after twelve weeks and found the maximum effects only at the second measurement point. The most important meta-analysis therefore only admitted studies that ran for at least twelve weeks. Short self-experiments over a few days end before the studies even begin to measure.

Q: Does Bacopa only work on memory?

According to the data, essentially yes. The review by Pase found improvements in 9 of 17 tests of free recall and hardly any indications in other domains. The 2026 network meta-analysis found the effect in working and short-term memory and explicitly no significant differences in attention and processing speed. An older meta-analysis, by contrast, had located the effect in reaction times, and this part was not reproduced later.

Q: Why does Bacopa give me stomach problems?

Because Bacopa has a cholinergic effect and thereby increases the tone and peristalsis of the stomach and intestines. This is the most frequently documented side effect. In an Australian study, increased stool frequency, abdominal cramps and nausea under Bacopa compared with placebo were explicitly reported. In a 30-day phase I study, the complaints subsided on their own.

Q: Does Bacopa affect the thyroid?

There is exactly one relevant finding on this, and it comes from an animal experiment. In male mice, a leaf extract of 200 milligrams per kilogram raised the T4 level by 41 percent, while T3 remained unchanged. Reviews derive from this a caution in hyperthyroidism and under thyroid medication. There are no human data on thyroid function under Bacopa.

Q: Is Bacopa better than ginkgo?

A 2026 network meta-analysis compared both in a joint model for the first time, across 29 randomized trials and 2,107 participants. High-dose Bacopa performed better in working memory than high- and low-dose ginkgo and than placebo. However, the authors themselves note that there are no direct head-to-head studies and that the informative value is therefore limited.

Q: Does Bacopa help with dementia or Alzheimer’s?

No, there is no evidence for that. A 2022 systematic review found five randomized trials in Alzheimer’s dementia, rated all five as at high risk of bias and concluded that the certainty of evidence was very low. There was no difference from placebo and none from donepezil. The available positive data come from studies in healthy adults and older people without dementia.

Adaptogen · Brahmi

Bacopa monnieri, called Brahmi in India, is one of the few nootropics for which genuine meta-analyses exist. The finding is surprisingly narrow: the effect appears almost exclusively in a single cognitive domain, and it only appears after weeks.

In short

Bacopa is an extract from an Ayurvedic marsh herb, standardized to saponins called bacosides. Several meta-analyses find an effect on memory, more precisely on the delayed and free recall of previously learned material. The effect is small, it only shows up in the studies after around twelve weeks of daily intake, and at the level of individual tests it is poorly reproducible. The largest analysis pools 29 randomized trials with 2,107 participants and sees a clear lead over placebo and ginkgo in working memory. The catches: gastrointestinal complaints are well documented, and a good share of the positive papers come from the same region of the world and use the same commercial extract.

What Bacopa is

Bacopa monnieri is a low-growing marsh herb from India, Southeast Asia and Australia. In the Ayurvedic tradition it is one of the Medhya Rasayana, the remedies for the mind and memory. What is sold today is not the plant, but an extract standardized to triterpene saponins: bacosides A and B.

This standardization is the heart of the matter. The best-studied extract is called CDRI 08, is adjusted to no less than 55 percent bacosides and comes from the Central Drug Research Institute in Lucknow, India; since 2009 an Australian company has been selling it as KeenMind. Anyone buying Bacopa is therefore buying a particular extraction method.

The one domain

If you lay the reviews side by side, it is striking how narrow the finding is. In 2012 Pase and colleagues evaluated six randomized trials, all over twelve weeks, with three extracts and daily doses of 300 to 450 milligrams. Bacopa improved performance in 9 of 17 tests of free recall. In all other domains the authors found hardly any indications.

The newest and largest paper confirms this narrowness. A 2026 network meta-analysis pools 29 randomized trials with 2,107 healthy adults. High-dose Brahmi from 600 milligrams daily performed better than placebo in working memory with a standardized mean difference of 2.03, confidence interval 1.28 to 2.78. The lead was also significant for short-term memory. For attention and processing speed, by contrast, no significant differences were found.

This is remarkable because the older and most-cited meta-analysis had its finding precisely there. In 2014 Kongkeaw and colleagues found a time in the Trail B test shortened by 17.9 milliseconds and a choice reaction time shortened by 10.6 milliseconds, both with p below 0.001, and attributed this to attention speed. The larger analysis does not reproduce exactly this part. What remains across all three papers is memory.

Why it takes weeks

Bacopa is not a substance where you notice something the same day. In 2001 the first large controlled study by Stough and colleagues tested at baseline, after five and after twelve weeks and found the maximum effects only at the last measurement point. The time window has become established as a convention: Kongkeaw only admitted studies of twelve weeks or more, and in Pase all six ran for exactly twelve weeks. Short self-experiments therefore end before the studies even begin to measure.

The most likely reason lies in the mechanism. A Thai study in 60 healthy older people measured, after twelve weeks, a suppression of acetylcholinesterase activity in plasma and shortened N100 and P300 latencies in the EEG. The target structure has thus been confirmed in humans, not only in animal models. The chain of action up to better recall, however, has not been quantified.

Few places, one extract

The meta-analysis by Kongkeaw is based on nine studies with 518 participants, of whom 437 were analyzed. If you look at where they were conducted, the base becomes narrow: four come from Australia, namely Stough 2001 in Victoria, Roodenrys 2002 in Wollongong, Stough 2008 at Swinburne University and Morgan and Stevens 2010 in Lismore. The two Stough papers come from the same group and use the same extract, KeenMind or CDRI 08.

This does not invalidate the results, but it limits them: an effect found several times by the same group with the same product is less independently confirmed than the number of studies suggests. This is also where the most uncomfortable individual finding comes in. A 2021 quantitative evaluation of the clinical data concluded that no two studies found significant changes in the same two tests. If every study runs a dozen tests and each finds something in different ones, some of the hits are chance.

What is well supported

The most stable finding is delayed recall. It appears in the methodologically strongest reviews, in individual studies at different locations and across several extracts. In Morgan and Stevens, Bacopa improved the delayed recall trial of the Rey test with p equal to 0.001 and total learning with p equal to 0.011, while the figure test and the Trail Making Test remained unremarkable. In Roodenrys, the learning rate remained unchanged, while the rate of forgetting newly learned information slowed. So Bacopa does not make you learn faster, but makes what you have learned stick longer.

What the studies show

Kongkeaw 2014, the reference meta-analysis

Meta-analysis of randomized, placebo-controlled trials with standardized extracts, at least twelve weeks of administration and no concomitant medication, search up to June 2013. Nine studies with 518 participants met the criteria, 437 were included in the analysis, and the risk of bias was considered low. Time in the Trail B test shortened by 17.9 milliseconds, confidence interval minus 24.6 to minus 11.2; choice reaction time by 10.6 milliseconds, interval minus 12.1 to minus 9.2. Both with p below 0.001.

Tiemtad 2026, Bacopa versus ginkgo

Network meta-analysis, search in November 2024, 29 randomized trials with 2,107 healthy adults, risk of bias according to Cochrane RoB 2. High-dose Brahmi from 600 milligrams daily improved working memory compared with placebo with a standardized mean difference of 2.03, compared with the lower dose of 300 to under 600 milligrams with 1.84 and compared with high-dose ginkgo from 240 milligrams with 1.94, with a SUCRA rank of 100 percent. Direct head-to-head studies are lacking.

Calabrese 2008, the cleanest single trial

Randomized, double-blind, placebo-controlled trial in Portland, Oregon, with a six-week placebo run-in and twelve weeks of treatment. 54 participants aged 65 and over, mean age 73.5 years, were randomized, and 48 completed the study, 24 per group, with 300 milligrams of dry extract daily. The primary endpoint was delayed recall in the Rey test, and it was met. Divided attention task, mood and blood pressure showed no effect.

Sathyanarayanan 2013, the counter-finding

Randomized, double-blind, placebo-controlled parallel-group study in Bangalore with 72 healthy adults aged 35 to 60, who received 450 milligrams of extract or placebo over twelve weeks. Verbal learning, memory, attention and anxiety were measured. Result: no significant differences in any cognitive measure, only a trend toward lower state anxiety. The primary endpoint was missed.

Where the data stop

The meta-analyses do not agree on which domain is affected at all: Kongkeaw saw attention speed, Tiemtad working and short-term memory, Pase free recall. A 2026 systematic review with seven studies found no benefit of Bacopa at all in healthy people, neither on speed nor on attention, working memory, language or overall performance. This is a serious contradiction in the literature, not a marginal finding.

Status, approval and legal

In Germany, Bacopa monnieri is not approved as a medicine, but is considered an ingredient of food supplements. This has a surprising consequence: legally, such ingredients must not exert any pharmacological effect, so a medicinal plant may only be used in an amount at which the expected effect does not occur. There are no binding maximum amounts, only the general EU maximum levels for contaminants. There is also no authorized health claim: EFSA lists 1,548 botanical claims that are on hold under Article 13. With Ayurvedic products, quality must also be taken into account. In 2016 the Lower Saxony State Office for Consumer Protection found lead up to 12.7, mercury up to 87.5 and arsenic up to 7.0 milligrams per kilogram in three such products; in 2018 the samples were unremarkable. This concerns the product category, not the extract.

Safety

The most common side effect concerns the gastrointestinal tract, and unlike with many plant substances, this is well documented in studies here. Morgan and Stevens report increased stool frequency, abdominal cramps and nausea under Bacopa compared with placebo; in Calabrese it was mainly upset stomach, in 9 people on Bacopa and 10 on placebo. The reason lies in the mechanism: the cholinergic effect increases the tone and peristalsis of the stomach and intestines. For the same reason, caution applies with a tendency to a slow pulse, with asthma and COPD, with stomach ulcers and with urinary tract obstruction; the combination with acetylcholinesterase inhibitors such as donepezil belongs in a doctor’s hands. On the thyroid, the situation must be stated clearly: one animal finding, no human data. In male mice, a leaf extract of 200 milligrams per kilogram raised the T4 level by 41 percent, T3 remained unaffected; from this, reviews derive caution in hyperthyroidism and under thyroid medication. This has never been tested in humans. In addition, there are drug interactions, because Bacopa inhibits several cytochrome isoenzymes. An evaluation of 1,816 adverse event reports filtered out 30 reports with a high probability of causality, including Bacopa with agomelatine and moclobemide. For pregnancy and breastfeeding the data are insufficient. The frequently cited advice to stop Bacopa before surgery is based on the mechanism, not supported by case reports.

BK-Score Supported, with caveats

Human evidence		6
Mechanism		6
Safety data		6
Hype gap		5
Track record of use		7

One of the few nootropics with several meta-analyses. The largest, a 2026 network meta-analysis of 29 RCTs with 2,107 healthy adults, finds an advantage in working memory compared with placebo (SMD 2.03; 95 % CI 1.28–2.78) and explicitly none in attention and processing speed. It thus does not reproduce the speed finding of the older meta-analysis by Kongkeaw (9 RCTs, 518 participants: Trail B −17.9 ms, choice reaction time 10.6 ms, both p < 0.001). What holds across the papers is a single domain: memory, more precisely free and delayed recall (Pase 2012: 9 of 17 tests). The effect takes around twelve weeks, little is repeated at the test level, and a good share of the positive studies come from Australia, some from the same group with the same extract. Acetylcholinesterase inhibition has been measured in plasma in humans; long-term data are lacking.

The score rates the state of knowledge, not the substance. “Safety data 9” means well studied – not harmless.
Subjective assessment by Biohacking Kompakt based on published scoring rules – not a scientific rating and not a medical recommendation. Rules and all ratings (German)

Frequently asked questions about Bacopa Monnieri

How long does it take for Bacopa to work?