Alpha-GPC (L-alpha-glycerylphosphorylcholine) is sold for focus, memory and workout power, and it shows up in many nootropic stacks and pre-workouts. In 2021 a study of about 12 million South Korean adults linked it to a higher risk of stroke, and that study is now cited in many online warnings about it.
Both stories need context. The stroke study is large but observational, and the people prescribed alpha-GPC in Korea were older and sicker than those who were not. The trials behind the focus and power claims are small, short and mostly paid for by ingredient makers. Below we go through both, with the numbers, so you can judge for yourself.
The short version
- Our verdict: an unresolved safety signal and thin evidence of benefit. Neither the stroke risk nor the performance benefits are established.
- The stroke signal. Among 12 million Korean adults aged 50 and over, people prescribed alpha-GPC had about 43 percent higher stroke risk (hazard ratio 1.43) over about 9 years of follow-up than matched non-users, roughly 16 versus 11 strokes per 1,000 people per year.
- It cannot show cause. In Korea alpha-GPC was a prescription drug given for memory complaints and early signs of poor blood flow in the brain, so the users may already have been at higher risk. A second Korean study, in people with mild memory problems, did not find higher stroke risk.
- The benefit trials are tiny. Single-ingredient trials in healthy people enrolled 7 to 48 people, lasted one dose to two weeks, and were mostly funded by ingredient makers. Most reported a gain on one measure and no change on the others, and none we found tested months of use.
- It is a modest choline source. Alpha-GPC is about 40 percent choline by weight, so 300 mg gives about 120 mg of choline.
- Ask a doctor first if you are over 50 or have high blood pressure, heart disease, atrial fibrillation, diabetes, kidney disease, or a past stroke or mini-stroke.
What alpha-GPC is
Alpha-GPC is a small, water-soluble molecule the body makes when it breaks down phosphatidylcholine, a main fat in cell membranes. After you swallow it, enzymes in the gut split it into choline and glycerophosphate. The body uses choline to build membranes and to make acetylcholine, a chemical messenger in the brain and muscles. Whether extra choline from a supplement changes acetylcholine signaling in healthy people has not been shown.
Its legal status depends on the country. In South Korea it is a prescription drug, sold as choline alfoscerate under brands such as Gliatilin, and widely used: by 2019 almost half of Koreans newly treated for Alzheimer's disease were prescribed it alongside standard dementia drugs. It is a prescription medicine in some European countries too. In the US it is sold as a dietary supplement, including the branded forms AlphaSize and GeniusPure. A drug approval abroad does not show that a supplement works.
The Korean stroke study
The study behind the warnings was published in JAMA Network Open in November 2021 by researchers at Seoul National University. They used national health insurance records for 12,008,977 Korean adults aged 50 and over who had no history of stroke or mini-stroke (TIA) and had not taken dementia drugs. Anyone prescribed alpha-GPC for at least one day between 2006 and 2008 counted as a user, which came to 108,877 people. Each user was matched to 10 non-users of the same age, sex, income and overall illness score (the Charlson index, a standard tally of long-term health conditions), and the groups were followed from 2009 to early 2018.
| Strokes per 1,000 person-years | Alpha-GPC users | Non-users | Hazard ratio |
|---|---|---|---|
| Any stroke | 16.2 | 11.3 | 1.43 |
| Ischemic stroke (clot) | 9.6 | 7.1 | 1.34 |
| Hemorrhagic stroke (bleed) | 1.25 | 0.91 | 1.37 |
A person-year is one person followed for one year, so 16.2 per 1,000 person-years means about 16 strokes a year for every 1,000 users. A hazard ratio compares how often the event happened in each group over time: 1.43 means users had strokes about 1.43 times as often as matched non-users, or about 43 percent more often. Any stroke also counts other and unspecified stroke types and related brain blood-vessel codes, so the two rows below it do not add up to it. Rates are our arithmetic from the paper's event counts and follow-up time.
In plain terms, users had about 5 more strokes per 1,000 people per year than matched non-users. Before matching, the raw gap was much bigger (16.2 versus 6.9), which shows how much of the difference came from age and illness. The link held up in two kinds of checks. It remained after also adjusting for smoking, drinking, exercise, weight, blood pressure, diabetes and cholesterol in a subgroup with health-screening data. It also remained after ignoring strokes in the first one to four years of follow-up, though it shrank a little with each year left out.
The paper calls the pattern dose-dependent, but no dose in milligrams was recorded. What rose was the length of the prescription. Compared with people prescribed it for under 2 months, the hazard ratio was 1.13 for 2 to 6 months, 1.18 for 6 to 12 months and 1.36 for more than 12 months. In September 2026 the journal corrected two stroke codes listed in the methods and an author credit; the correction notice lists no changes to the results.
Why it cannot show cause
A records study can show that two things go together, not that one causes the other. The biggest problem here is what researchers call confounding by indication: the reason someone got the drug may itself raise their risk. In Korea alpha-GPC was approved for three uses: "secondary symptoms of cerebrovascular defects" and degenerative brain conditions such as dementia (memory and thinking problems blamed on poor blood flow or brain aging), mood and behavior changes, and depression in older age. The authors note this themselves and write that users "may already have subclinical atherosclerotic changes," meaning hardened or narrowed arteries that are not yet causing symptoms.
The numbers fit that picture. Before matching, users averaged 68 years old against 62 for non-users, and two-thirds had a high illness score against under a third of non-users. Matching on age, sex, income and a general illness score cannot capture a specific complaint such as memory loss or confusion from early blood-vessel disease, which may be what many prescriptions were for. The study also had no information on use after 2008 and did not account for diet, kidney function or TMAO levels.
None of this means the signal is wrong. It means the study cannot tell apart two explanations: alpha-GPC raising stroke risk, or alpha-GPC being given to people already on the way to a stroke.
The second Korean study
A 2025 study in the Journal of Prevention of Alzheimer's Disease, mainly about dementia but also tracking strokes, looked at 508,107 Koreans newly diagnosed with mild cognitive impairment, 77 percent of whom were later prescribed alpha-GPC. Because almost everyone in that group had the same reason to take it, the indication problem is smaller. In its main analysis, users had a somewhat lower stroke risk; in a second, matched analysis there was no difference either way. The authors concluded that there "appears to be no reason to avoid" alpha-GPC "solely to prevent stroke."
This study has its own weaknesses. Anyone prescribed it even once counted as a user, the result depended on the analysis method, and its stroke-type labels appear to be swapped (the same coding mix-up the first study later corrected), so we do not quote its numbers by stroke type. Two Korean insurance-records studies point in different directions, and neither can show cause. The question is open.
The TMAO theory
The stroke study's authors proposed a mechanism they did not measure. Gut bacteria turn choline into trimethylamine (TMA), which the liver converts to trimethylamine N-oxide (TMAO), and higher blood TMAO has been linked in observational studies to heart attack and stroke. The human evidence connecting alpha-GPC to this is thin:
- In a crossover study of 6 healthy men, a single large dose of alpha-GPC (equal to 550 mg of choline, about 1,360 mg of alpha-GPC) raised TMAO, as choline salts did, while choline from egg phospholipids did not.
- In a study of 12 healthy men, funded by a drug company that employed the authors, a single dose did not raise TMAO significantly.
- In mice bred to develop artery plaque, alpha-GPC at a very high dose for 16 weeks increased plaque, and TMAO tracked plaque size. Mouse results like this often do not carry over to people. Two of the authors also hold patents on heart-disease tests and treatments.
- A genetic study of TMAO itself did not find that people with naturally higher TMAO have more strokes.
We found no human study of long-term alpha-GPC use and TMAO, and none linking alpha-GPC to TMAO-related heart or brain events. For now, TMAO is a hypothesis worth testing, not an established explanation.
What the trials in healthy people found
Marketing for alpha-GPC leans on power, growth hormone and focus. Here are the single-ingredient trials in healthy people we found that tested power, growth hormone, mood or focus. Combination products are left out, because they cannot show what alpha-GPC did on its own. Two other small trials are also left out because they did not test everyday performance: one found faster heart rate and blood pressure recovery after sprints in 12 women after a single 1,000 mg dose, and one found that 10 days of alpha-GPC blunted the attention and memory loss caused by the drug scopolamine in 32 young volunteers.
| Trial | Who and how long | Dose | What it found |
|---|---|---|---|
| Ziegenfuss 2008 (conference poster) | 7 trained men, one dose | 600 mg | Bench press force about 14 percent higher; growth hormone rose more, but started lower that day |
| Kawamura 2012 | 8 men, one dose | 1,000 mg | Growth hormone higher at 60 minutes; no performance test |
| Bellar 2015 | 13 college men, 6 days | 600 mg a day | Lower-body force rose more than on placebo (borderline); upper-body force not significantly different |
| Parker 2015 (conference poster) | 20 young adults, one dose | 200 or 400 mg | No significant effect on mood, thinking or performance |
| Marcus 2017 | 48 college men, 7 days | 250 or 500 mg a day | No difference in strength or a reaction-time test; jump power best at 250 mg, not 500 mg |
| Tamura 2021 | 39 adults, 2 weeks, single-blind | 400 mg a day | A trend toward more motivation, significant only at night; no effect on anxiety |
| Kerksick 2024 (GeniusPure) | 20 trained men, one dose | 315 or 630 mg | Better on one thinking test (Stroop) of three; no change in performance or growth hormone |
Every one of these was small, and none ran longer than two weeks. Most were funded by companies selling alpha-GPC ingredients: Chemi Nutra (AlphaSize) supported the 2008, 2015 and 2017 sports studies, and NNB Nutrition (GeniusPure) sponsored the 2024 trial. Two of the trials are conference posters; we found no full paper for either. A rise in growth hormone after one dose does not by itself mean more muscle or strength, and we found no trial that tested it. We found no review pooling the healthy-adult trials.
Overall, most of these small trials found a short-term gain on one measure, such as force, growth hormone or one thinking test, and no change on the others; one found nothing, and none we found tested what happens over months. That is not enough to say alpha-GPC improves power or focus.
Dementia research is drug research
Much of the published alpha-GPC research is in people with Alzheimer's disease, vascular dementia or recent strokes, using the prescription drug at about 1,200 mg a day, sometimes by injection. A 2003 trial of 261 people with Alzheimer's reported better scores on cognitive tests over six months, and an Italian trial that added it to donepezil has published interim results. Not every trial was positive: a one-year trial in 62 people with small-vessel brain disease, which tested it together with the drug nimodipine, found no difference. Reviews of these trials come mostly from the same Italian research group that ran several of them.
These are drug studies in diagnosed disease, at higher doses than supplements, and they do not show what a capsule does for a healthy person. Alpha-GPC is not a treatment for dementia or stroke. New memory problems need a medical evaluation, and stroke symptoms such as sudden weakness, face drooping or trouble speaking need emergency care.
What regulators have done
South Korea. Korean press reports describe a long review of the drug. In 2020 the government sharply raised what patients pay for it outside dementia care and ordered makers of 255 products to run new clinical trials proving it works. In 2021 two of its three approved uses were removed from labels. In 2026 Korean trade press reported that the first of the required trials, in mild cognitive impairment, missed its main goal, with a final decision pending at the time of the report. These actions were about whether it works, not about stroke, and we found no Korean regulatory action on stroke risk.
United States. Alpha-GPC is sold as a supplement and is not approved to treat any disease. In 2012 Chemi Nutra notified the FDA that it considered alpha-GPC safe as a choline source in foods at 10 to 100 mg per serving, and the FDA had no questions. In 2024 another company filed a similar notice for up to 600 mg per serving in drinks, bars and powders. The FDA wrote that there "was not sufficient data to support the safety conclusion," and the review ended at the company's request, after the FDA recommended that step. That letter did not mention stroke.
Alpha-GPC dosage and choline content
Trials in healthy people used 250 to 600 mg a day for up to two weeks, or single doses of 200 to 1,000 mg. Supplements usually supply 300 to 600 mg a day. Check the label for the amount of active alpha-GPC: some products list 600 mg of a "50%" powder, which is 300 mg of alpha-GPC. Our alpha-GPC supplement ranking compares five products on dose, purity, testing and cost.
By weight, alpha-GPC is about 40.5 percent choline, about the same as choline bitartrate, which is about 41 percent. So 300 mg gives about 120 mg of choline and 600 mg about 243 mg, against a daily Adequate Intake of 425 mg for women and 550 mg for men. The choline itself is not the safety question: the adult upper limit is 3,500 mg of choline a day. If you mainly want choline, eggs or a choline salt cost less. Our choline supplement ranking and alpha-GPC vs citicoline comparison cover the forms.
Side effects and who should be careful
The short trials in healthy people reported few side effects, but they were too small and brief to pick up rare or long-term problems. In a large open study of stroke patients given the drug by injection and then by mouth for six months, about 2 percent reported side effects, mainly heartburn, nausea, sleeplessness or restlessness, and headache. One 2017 trial found lower thyroid-stimulating hormone (TSH), a pituitary hormone that controls the thyroid, after 500 mg; we found no other study that checked this. We found no study of long-term safety in healthy people. In the Korean study the link was, if anything, stronger in people aged 50 to 64 than in those 65 and older (hazard ratio 1.79 versus 1.52), but no one under 50 was included, so nothing yet tells us whether the association, if it is causal, would apply to younger, healthy users.
- Higher stroke or heart risk. If you are over 50, or have high blood pressure, heart disease, atrial fibrillation, diabetes, kidney disease, or a past stroke or TIA, talk with your doctor before taking it daily.
- Not a treatment. Alpha-GPC does not treat dementia or stroke. Sudden weakness, face drooping or trouble speaking needs emergency care.
- Prescription medicines. If you take dementia drugs such as donepezil, anticholinergic medicines, or Parkinson's medicines, ask your doctor first and do not change those medicines on your own.
- Pregnancy, breastfeeding and children. We found no safety data; get choline from food or a prenatal vitamin instead.
Frequently asked questions
Does alpha-GPC cause strokes?
That has not been shown. A 2021 study of about 12 million Korean adults aged 50 and over found that people prescribed alpha-GPC had about 43 percent higher stroke risk over about 9 years of follow-up than matched non-users, roughly 16 versus 11 strokes per 1,000 people per year. In Korea it was prescribed for early blood-vessel and memory complaints, so users may already have been at higher risk, and a 2025 Korean study in people with mild memory problems did not find higher stroke risk. The question is open, so people over 50 or with heart or stroke risk factors should ask a doctor before taking it daily.
What is alpha-GPC good for?
The evidence for benefits in healthy people is thin. Small trials of one dose to two weeks, with 7 to 48 people each and mostly funded by ingredient makers, mostly found a short-term gain on one measure, such as lower-body or bench press force, a rise in growth hormone after one dose, or a better score on one thinking test, with no change on the other measures; one found no effect at all. We found no trial that tested whether it improves strength, muscle or focus over months.
Is alpha-GPC the same as choline alfoscerate?
Yes. Choline alfoscerate is the drug name for alpha-GPC (L-alpha-glycerylphosphorylcholine). It is a prescription drug in South Korea and some European countries and a dietary supplement in the US, where AlphaSize and GeniusPure are branded forms.
How much choline is in alpha-GPC?
Alpha-GPC is about 40.5 percent choline by weight, so 300 mg supplies about 120 mg of choline and 600 mg about 243 mg. The daily Adequate Intake is 425 mg for women and 550 mg for men. Some labels list 600 mg of a 50 percent powder, which is 300 mg of alpha-GPC.
How much alpha-GPC do studies use?
Trials in healthy adults used 250 to 600 mg a day for up to two weeks, or single doses of 200 to 1,000 mg. Drug trials in people with dementia used about 1,200 mg a day for six months or longer, which is a prescription-drug use, not a supplement dose. This is not dosing advice.
Is alpha-GPC safe?
Its long-term safety has not been established. The short trials we found in healthy people, none longer than two weeks, reported few side effects, but they were too small and brief to detect rare problems. A study of 12 million Korean adults aged 50 and over linked prescription alpha-GPC to about 43 percent more strokes. It cannot show that alpha-GPC caused them, but the signal has not been ruled out. If you are over 50 or have high blood pressure, heart disease, atrial fibrillation, diabetes, kidney disease, or a past stroke or TIA, talk with your doctor before taking it daily.
What are the side effects of alpha-GPC?
In a large open study of stroke patients, about 2 percent reported side effects, mainly heartburn, nausea, trouble sleeping or restlessness, and headache. Short trials in healthy people reported few problems, but we found no study of long-term safety in healthy people. Ask your doctor first if you take dementia, anticholinergic or Parkinson's medicines, and avoid it in pregnancy, breastfeeding and children, where we found no safety data.
Is alpha-GPC better than citicoline?
Neither has strong evidence in healthy people. Alpha-GPC carries more choline per milligram (about 40 percent versus about 20 percent for citicoline), while citicoline has larger drug trials, most of which found no benefit, and no stroke signal of the kind reported for alpha-GPC. Our citicoline evidence review and alpha-GPC vs citicoline comparison cover the details.
The bottom line
The stroke concern comes from a very large Korean records study that found more strokes among people prescribed alpha-GPC, about 5 more per 1,000 people per year. That study cannot separate the drug from the reasons people were given it, and a second Korean study did not find the same. The TMAO explanation rests on two tiny human studies that disagree and a mouse experiment. The safety question is open.
The benefit side is weak too: a handful of small, short, mostly company-funded trials with mixed results for power, growth hormone and focus. If you try it anyway, trials in healthy adults used 250 to 600 mg a day for no more than two weeks, so expect little and do not assume that months of daily use have been tested. If you are over 50 or have stroke or heart risk factors, the uncertainty weighs more heavily, so talk with your doctor first.