Does Alpha-GPC Work — and What About the Stroke-Risk Study?

By Erin Rose · Updated July 18, 2026 · Methodology

Educational overview — not medical advice. Alpha-GPC (choline alfoscerate) is a prescription drug abroad for dementia and post-stroke cognitive decline; this page discusses that evidence for context, not as a treatment recommendation. This statement has not been evaluated by the FDA. This product is not intended to diagnose, treat, cure, or prevent any disease.

Four tiers of evidence, and why conflating them is the whole problem

Alpha-GPC marketing routinely blends genuinely different kinds of evidence into one confident-sounding "it works" claim. Separated honestly, they tell four different stories.

Tier 1: The mechanism (real, and confirmed in humans)

Alpha-GPC is a choline-donor compound. A small pharmacokinetic trial found that a single 1,000mg intramuscular dose produced a rapid, substantial rise in plasma free choline — peaking at 15–30 minutes, and higher than a comparable dose of citicoline (Gatti 1992). This confirms alpha-GPC reliably delivers choline into the bloodstream. It does not, on its own, establish a cognitive or performance outcome — a rising blood marker is not the same as a proven clinical benefit, and every claim built on top of this mechanism needs its own separate evidence.

Tier 2: Disease-population, prescription-dose evidence (real, but doesn't transfer to healthy users)

Abroad, alpha-GPC is an approved prescription drug (choline alfoscerate, brand name Gliatilin) in Italy, Russia, and other countries for dementia and post-stroke cognitive decline. A 261-patient multicenter RCT in mild-to-moderate Alzheimer's dementia found significant ADAS-Cog improvement over 180 days at 1,200mg/day — 400mg three times daily (De Jesus Moreno Moreno 2003). A 2023 systematic review of 8 studies (7 RCTs plus 1 cohort) in dementia and cerebrovascular cognitive-impairment populations found choline alphoscerate, alone or combined with donepezil, improved cognition, behavior, and functional outcomes in those disease populations (Sagaro 2023). And a large, open-label (not placebo-controlled) 1994 Italian trial in 2,044 patients with recent stroke or TIA found Mini-Mental State scores improved from 21 to 24.3 over 6 months, with adverse events in only 2.14% of patients — a real safety-volume dataset, but a weaker-design efficacy signal given the absence of a placebo arm (Barbagallo Sangiorgi 1994). This tier is genuinely strong as clinical evidence in the population it was tested in. It is, without exception, disease-population evidence at roughly 2–4× a typical retail dose. None of these trials say anything about a healthy adult taking 300mg for daily focus, and none of them should be cited as if they did.

Tier 3: Healthy-adult sports-performance evidence (small, real, inconsistent)

Three trials tested alpha-GPC in healthy, mostly college-age or recreationally-trained men at 250–630mg doses. Bellar 2015 (n=13) found lower- but not upper-body strength gains after 6 days at 600mg/day. Marcus 2017 (n=48) found a non-dose-dependent jump-power result — the lower 250mg dose beat the higher 500mg dose, directly undercutting a "more is better" reading. Kerksick 2024 (n=20) found a single significant Stroop-test improvement at 630mg, alongside three null cognitive tests and null physical-performance and growth-hormone results in the same trial; its lead author is a paid scientific advisor to the study's supplement sponsor, NNB Nutrition. See the full breakdown on the exercise-performance page. This tier is real but thin, and the claimed "growth hormone spike" (traced mainly to an unindexed 2008 conference abstract) has no proven downstream benefit in any indexed trial — Kerksick 2024, the best available data, found no GH difference at all.

Tier 4: The safety signal (real, large, and the one most often omitted)

This is the tier that most changes the honest picture, and it's the one retail marketing for alpha-GPC almost never mentions. Lee et al. (2021, JAMA Network Open, PMID 34817582) used National Health Insurance Service claims data to follow 12,008,977 South Korean adults aged 50 and older, without prior stroke or Alzheimer's, for 10 years. Alpha-GPC use was associated with significantly higher total stroke risk — adjusted hazard ratio 1.46 (95% CI 1.43–1.48) — and risk rose in a dose-response pattern with cumulative days of use. This is observational data: it demonstrates an association, not proven causation, and the authors do not identify a confirmed biological mechanism connecting alpha-GPC to stroke risk. But it is one of the largest population-level studies ever conducted on this compound, published in a major peer-reviewed journal, and it is essentially never disclosed in retail marketing copy.

Why this matters more than the usual "more research needed"

Plenty of supplements have thin human evidence and get a generic "promising, needs more research" writeup. Alpha-GPC is a sharper case: it has real prescription-drug-grade evidence, just not for the population most retail buyers are in, and it has a genuine, large-scale safety signal that most competitors in the AI-search results for this ingredient simply don't cite. That's a materially different, more nuanced situation than "nobody has looked yet." It doesn't mean alpha-GPC is dangerous for everyone, and it doesn't mean the sports-performance signal is fake. It means the honest framing has to hold the mechanism, the disease-population pedigree, the modest healthy-adult signal, and the stroke-risk association together — not cherry-pick whichever one supports a sale.

Frequently asked questions

Related

• Alpha-GPC: what it is & who it's for
• Alpha-GPC dosage guide — trial dose vs prescription dose vs retail
• Best Alpha-GPC — ranked by actual-mg disclosure, then cost per mg
• Alpha-GPC for power, strength & focus during exercise
• Phosphatidylserine — another cognition supplement with a discontinued-raw-material honesty gap
• Citicoline vs. Alpha-GPC — the other choline-donor nootropic: no comparable stroke-risk signal, but it got (and failed to beat placebo in) the prospective stroke RCT alpha-GPC never faced

Sources