Alpha-GPC Dosage: What the Trials Actually Used

Nearly every clinical alpha-GPC trial used 1,200 mg per day, given as 400 mg three times daily, and the participants were people with mild-to-moderate Alzheimer’s or vascular dementia — often alongside donepezil.¹ Sports and acute-cognition studies use 300–600 mg. There is no convincing healthy-adult cognitive data at any dose, and the compound carries the most serious safety signal in this reference: a Korean national cohort of more than 12 million adults aged 50+ found adjusted hazard ratios of 1.43 for total stroke among alpha-GPC users over ten years, with a dose-response gradient.²

What a disclosed label lets you check

Neuro-Thrive

It lists alpha-GPC at 150 mg — one-eighth of the 1,200 mg used in the clinical trials, and below the 300–600 mg acute-study range too. It is on our approved list because it publishes every dose on the panel, including bacopa monnieri at 300 mg standardized to 50% bacosides, which is a genuine trial dose. The caveats are not optional: its PQQ is 10 mg where the cognitive trial that ingredient is sold on used 20 mg/day, and the alpha-GPC in it is the same compound the Korean cohort associated with roughly 43% higher 10-year stroke risk. If you have cerebrovascular history, uncontrolled hypertension, or take an anticoagulant, this is a clinician conversation before it is a purchase.

We earn a commission if you buy through this link. It costs you nothing extra and it does not change what we write.

Full Neuro-Thrive review: doses, price, guarantee →

Check the current price →

Written by the Fog & Clarity editorial team · Published 8 August 2026 · Last updated 8 August 2026

The doses that appear in trials

Alpha-GPC — L-alpha-glycerylphosphorylcholine, also called choline alfoscerate, and a prescription product (Gliatilin, Delecit) in Italy and Korea — has an unusually tidy dose literature. Almost all of it converges on one number, and almost all of it was collected in a population that is not the one buying it in capsules.

Source	Population	Daily dose	Duration	Outcome
Sagaro 2023 meta-analysis, 8 studies	861 participants, clinical dementia	1,200 mg typical	Varies by trial	Monotherapy MMSE MD 3.50 (95% CI 0.36–6.63)
Sagaro, donepezil add-on arm	Dementia patients on donepezil	1,200 mg typical	Varies by trial	MMSE MD 1.72; ADAS-Cog MD −5.76
Standard clinical protocol	Dementia populations	1,200 mg (400 mg × 3)	Varies by trial	The dose nearly all trials used
Sports and acute-cognition studies	Healthy adults	300–600 mg	Acute	No convincing cognitive data
Korean national cohort	108,877 users, adults 50+	Prescription use	10 years	aHR 1.43 total stroke, dose-response

The 2023 systematic review and meta-analysis by Sagaro and colleagues pooled 8 studies and 861 participants across Russia, Italy and Mexico.¹ The monotherapy MMSE mean difference of 3.50 is clinically non-trivial in dementia, which is where it was measured. The evidence base is also heavily concentrated in Italian research groups, largely through the ASCOMALVA work, and that concentration is a real limitation of the literature rather than a criticism of any one trial.

What that means for a healthy adult

Evidence tier for alpha-GPC on this site: Moderate in clinical dementia populations, None in healthy adults — with a safety signal attached to both.

Everything positive in the alpha-GPC file was measured in patients. Extending it to a healthy 35-year-old who wants sharper focus is an extrapolation, not a finding, and the Alzheimer’s Drug Discovery Foundation’s read is the same.³ The sports literature at 300–600 mg is about power output rather than cognition, and it has not produced convincing cognitive results.

There is a second reason to be careful about that extrapolation, and it is not about efficacy. The trial populations were supervised, prescribed, and monitored. A person buying alpha-GPC in a blend is none of those things, and the risk data below was collected in people over 50 with cognitive complaints — a group whose vascular risk is elevated to begin with.

The stroke signal, and how seriously to take it

⚠️ This is the most important number on the page

A population-based retrospective cohort using South Korea’s National Health Insurance Service database — more than 12 million adults aged 50 and over, comparing 108,877 alpha-GPC users against 11.9 million non-users across a ten-year follow-up from 2009 to 2018 — found adjusted hazard ratios of 1.43 for total stroke, 1.34 for ischemic stroke and 1.37 for hemorrhagic stroke, with a dose-response relationship. Anyone with cerebrovascular disease, a prior stroke or TIA, uncontrolled hypertension, or an anticoagulant prescription should not take alpha-GPC without a clinician’s involvement.

Now the honest counterweights, because a page that reports only the alarming half is doing the same thing the marketing does in the other direction. The study is observational, and confounding by indication is a genuine concern: in Korea, the people prescribed alpha-GPC are by definition people with cognitive complaints, and that group carries higher vascular risk before anyone dispenses anything. A separate Korean analysis has suggested delayed dementia conversion in alpha-GPC users, so the literature is genuinely conflicted rather than uniformly negative.⁴ Both results should be read together; neither belongs on a page alone.

What tips our editorial judgment is the size of the cohort and the dose-response gradient, which is the pattern you would expect if the association were causal. There is also a mechanistic hypothesis worth knowing: dietary choline raises TMAO, a proposed mediator of atherosclerosis. That is a hypothesis, not a demonstrated pathway. We cover the cohort itself in more detail on its own page.

Why the dose on your bottle is probably different

Almost no consumer supplement contains 1,200 mg of alpha-GPC, and for once that gap is not straightforwardly a scandal — it is a formulation choice sitting on top of an unresolved risk question.

Blend totals. Alpha-GPC is typically included at 50–300 mg inside a multi-ingredient formula, because 1,200 mg would consume the whole capsule. That is the arithmetic proprietary blends are designed to hide when the total is undisclosed.
Hygroscopic material. Alpha-GPC absorbs water aggressively, which is why it is often listed as a 50% powder on a carrier. A “300 mg alpha-GPC 50%” line is 150 mg of the active compound.
Serving splits. As always, read the serving size before the milligram number.

Among the products we are willing to link, Neuro-Thrive discloses alpha-GPC at 150 mg, and disclosure is precisely what lets us tell you that this is one-eighth of the clinical dose and that the same ingredient is the one behind the stroke cohort. Its bacopa at 300 mg standardized to 50% bacosides is a real trial dose;⁸ its PQQ at 10 mg is half the 20 mg/day used in the trial that ingredient is sold on.⁷ A published panel is the precondition for a real argument, not a substitute for one — and if the alpha-GPC content is the part that worries you, that is a legitimate reason to skip it.

Form matters more than milligrams here

Alpha-GPC, citicoline and choline bitartrate are three different molecules with three different evidence files, and treating them as interchangeable “choline” is the single most common error in this category. Citicoline’s evidence sits in age-related and vascular cognitive impairment plus small acute attention effects in lower-performing individuals, and it is among the best-tolerated agents in this reference with no serious safety issues reported, including in use up to three years.⁵ Alpha-GPC’s evidence is larger in effect size but confined to dementia populations, and it comes with the cohort data above.

If you are choosing on risk rather than on effect size, that asymmetry decides it, and our citicoline dosage page covers the alternative in detail.

Safety ceiling and who should not take this

The reported adverse-effect list is longer than most supplements in this category: constipation, nervousness, heartburn, headache, insomnia, dizziness, rash and confusion. None of those is the reason to be cautious. The cohort data is.

⚠️ Talk to your prescriber first

Alpha-GPC is additive with cholinesterase inhibitors (donepezil, rivastigmine, galantamine) — that combination appears in the trials specifically because it is pharmacologically active, and it is not something to assemble yourself. Anyone with cerebrovascular disease, a prior stroke or TIA, uncontrolled hypertension, or on anticoagulants should avoid it pending better data. If a surgery is scheduled, apply the general rule and stop supplements well in advance.

Huperzine A is worth mentioning here only as a contrast: it is another cholinergic agent sold in the same blends, it is pharmacologically a reversible acetylcholinesterase inhibitor, and Cochrane concluded the evidence is inadequate to recommend it.⁶ Stacking two cholinergic compounds without supervision is exactly the scenario the interaction literature warns about. If a label contains both, our red-flag list explains why we treat that as disqualifying rather than impressive.

What we don’t know

Whether the stroke association is causal. An observational cohort with a dose-response gradient is suggestive and not decisive. No randomized trial has been powered for stroke as an outcome.
Whether 150 mg carries the same risk as 1,200 mg. A dose-response relationship implies lower doses carry less risk. It does not establish a threshold below which there is none.
Whether alpha-GPC does anything at all in healthy adults. Nobody has run the trial that would answer this. The healthy-adult claim is currently an extrapolation from patients.
How to reconcile the two Korean analyses. One suggests higher stroke risk, the other delayed dementia conversion, in overlapping populations. That is unresolved.

What would change our mind

A prospective randomized trial in adults aged 50+ with cardiovascular events as a pre-specified safety endpoint, large enough and long enough to detect a hazard ratio near 1.4 — or, failing that, an independent replication of the Korean cohort in a different health system with different prescribing patterns, which would separate the drug from the indication. If either came back null, alpha-GPC would move from “avoid pending data” to “a moderate-evidence option in supervised clinical use” on this site. If a replication confirmed the gradient, we would stop linking any formula containing it.

Frequently asked questions

Is 300 mg of alpha-GPC enough?

Enough for what is the question. It is the low end of the sports and acute-cognition range, and a quarter of the 1,200 mg used in the clinical trials that produced the cognitive results. Nothing in the literature supports 300 mg as a cognitive dose in healthy adults, because nothing in the literature supports any dose in healthy adults.

Is alpha-GPC or citicoline better?

They are answers to different questions. Alpha-GPC has the larger measured effect, in dementia populations, with a large observational stroke signal. Citicoline has a smaller and narrower effect with a much cleaner safety record over years of use. For someone healthy, buying on risk profile rather than effect size is the defensible choice.

Should I stop taking alpha-GPC because of the stroke study?

That is a decision to make with a clinician, not with a website — particularly if it was prescribed to you. What we can say is that the cohort is very large, the gradient is dose-dependent, and the counter-argument (confounding by indication) is real but does not dissolve the finding. If you have cerebrovascular history or take an anticoagulant, raise it at your next appointment rather than waiting for one.

Can I take alpha-GPC with donepezil?

Not on your own initiative. The combination appears in the clinical trials, which tells you it is pharmacologically active rather than harmless, and cholinergic effects are additive. That combination belongs to a prescriber managing both.

Does alpha-GPC help healthy people focus?

There is no convincing healthy-adult cognitive data at any dose. The marketing claim descends from dementia trials and from sports studies about power output. “No evidence exists” is the accurate answer, and it is the one most product pages will not give you.

Sources

These statements have not been evaluated by the Food and Drug Administration. Nothing here is intended to diagnose, treat, cure or prevent any disease.

This article is information, not medical advice. Talk to a doctor or pharmacist about your own situation, especially if you take prescription medication.