Choline sources — citicoline, alpha-GPC and dietary choline

Reviewed

Reviewed on 2026-09-19 against the evidence rules: every claim carries a grade, every claim and every safety flag carries a source, and every citation was re-resolved against the record it names. The read was delegated by the owner on 2026-09-18 and signed by the Lead on his behalf (D-295) — it is a delegated review, not the owner's own reading of the page. Every claim on it carries an evidence grade and a named weakest link. It is educational and is not medical advice.

Last reviewed 2026-08-23

One decent sponsored trial of citicoline in older adults with memory complaints, no adequate trial of alpha-GPC in healthy young adults, and a stroke association in twelve million people that the category does not mention.

Before anything else

Safety

Adverse effects
In a cohort of 12 million Korean adults aged 50 and over, people prescribed alpha-GPC had a 43% higher rate of stroke over the following ten years after matching on all measured covariates (adjusted hazard ratio 1.43, 95% confidence interval 1.41 to 1.46), and the risk rose with how long they took it. This is observational: people prescribed alpha-GPC are older and already concerned about memory, which is exactly the kind of confounding that inflates a hazard ratio. It is also the largest safety observation attached to anything on this site, and it is here rather than at the bottom of the page.
Source: Association of L-α Glycerylphosphorylcholine With Subsequent Stroke Risk After 10 Years
Product quality
Citicoline, alpha-GPC and choline bitartrate are sold beside each other as though they were interchangeable. They are different molecules with different evidence — the trial above is about citicoline at 500 mg and says nothing about the others.
Source: Citicoline and Memory Function in Healthy Older Adults: A Randomized, Double-Blind, Placebo-Controlled Clinical Trial; Association of L-α Glycerylphosphorylcholine With Subsequent Stroke Risk After 10 Years
Who should not
Choline requirements in pregnancy are a clinical matter with their own guidance, and nothing on this page is written for that situation.
Source: The relation of dietary choline to cognitive performance and white-matter hyperintensity in the Framingham Offspring Cohort
No data in your group
Every human result on this page is in adults aged 50 and over. For a reader in their twenties or thirties, all three of these compounds are untested against cognition.
Source: Citicoline and Memory Function in Healthy Older Adults: A Randomized, Double-Blind, Placebo-Controlled Clinical Trial; Association of L-α Glycerylphosphorylcholine With Subsequent Stroke Risk After 10 Years; The relation of dietary choline to cognitive performance and white-matter hyperintensity in the Framingham Offspring Cohort

Every claim, with its grade

What the evidence says, and how strong it is

A grade belongs to one claim, in one population, for one preparation — never to a compound, a food or a page. Open a grade to see the four links between the evidence and the thing you would actually do, and which of them is doing the least work.

In a trial of 100 healthy adults aged 50 to 85 with age-associated memory impairment, 12 weeks of citicoline improved episodic memory on a paired-associate test against placebo (0.15 against 0.06, p = 0.0025). Episodic memory was a secondary outcome in that trial, and the ingredient's manufacturer sponsored it.
In whom: Healthy adults 50–85 with age-associated memory impairment
What form: Citicoline (Cognizin) 500 mg/day for 12 weeks
Measured as: Episodic memory, reported as a secondary outcome
Evidence grade C: Early research suggests
Why this grade
Weakest link
Outcome — what was measured is not what you would notice. Episodic memory was a secondary outcome in a trial the ingredient's manufacturer sponsored, and a secondary outcome is a weaker thing than the one a trial was designed to test.
What was actually studied
Healthy adults 50–85 with age-associated memory impairment (n = 100)
Detail worth knowing
Citicoline (Cognizin) 500 mg/day for 12 weeks, 49 treated and 51 placebo. Episodic memory, assessed by the Paired Associate test, was a **secondary** outcome (0.15 against 0.06, p = 0.0025). NCT03369925. Cognizin is a Kyowa Hakko brand and two authors are affiliated with the sponsor.
The trial above recruited people aged 50 to 85 who had a memory complaint. No adequate trial in healthy adults under 50 was found on 2026-08-23, so there is nothing to grade for that group.
In whom: Healthy adults under 50 with no memory complaint
What form: Citicoline at any dose
Measured as: Any cognitive outcome
Evidence grade D: Human evidence insufficient
Why this grade
Weakest link
Population — the people studied are not the people reading. The one trial recruited people aged 50 to 85 who already had a memory complaint, so for a healthy adult under fifty there is no group here to reason from.
What was actually studied
Healthy adults 50–85 with age-associated memory impairment (n = 100)
Detail worth knowing
Citicoline (Cognizin) 500 mg/day for 12 weeks, 49 treated and 51 placebo. Episodic memory, assessed by the Paired Associate test, was a **secondary** outcome (0.15 against 0.06, p = 0.0025). NCT03369925. Cognizin is a Kyowa Hakko brand and two authors are affiliated with the sponsor.
No randomised trial adequate to grade a cognitive outcome for alpha-GPC in healthy young adults was found when this dossier was written on 2026-08-23. The compound's human record is largely in older clinical populations, and the largest observational study of it is about stroke rather than about memory.
In whom: Healthy young adults
What form: Alpha-GPC (L-alpha glycerylphosphorylcholine) at any dose
Measured as: Any cognitive outcome
Evidence grade D: Human evidence insufficient
Why this grade
Weakest link
Study quality — design, size, blinding or replication. No randomised trial adequate to grade a cognitive outcome was found on 2026-08-23; the compound's human record is in older clinical populations and its largest study is about stroke.
What was actually studied
Korean National Health Insurance members aged 50 and over without prior stroke or Alzheimer's disease (n = 12008977)
Detail worth knowing
After matching on all covariates, alpha-GPC users had adjusted hazard ratios of 1.43 (1.41–1.46) for total stroke, 1.34 for ischaemic and 1.37 for haemorrhagic stroke, rising with duration of use. Observational and confounded by indication; the authors call for work on the mechanism.
In the Framingham Offspring cohort, higher habitual dietary choline was associated with better cognitive performance. This is an association in an observational cohort using estimated intake, not a trial of a dose, and people who eat more choline differ from people who eat less in many other ways.
In whom: Framingham Offspring cohort participants
What form: Habitual dietary choline, estimated by food-frequency questionnaire
Measured as: Cognitive performance and white-matter appearance on MRI
Evidence grade C: Early research suggests
Why this grade
Weakest link
Preparation — what was tested is not what is sold. The exposure is habitual dietary choline estimated from a food-frequency questionnaire, which is not a dose anybody administered or verified.
What was actually studied
Framingham Offspring participants, dietary choline by food-frequency questionnaire
Detail worth knowing
An observational association between habitual dietary choline intake and cognitive performance. Association only; food-frequency intake is not a randomised dose.

Sources

Everything above, and where it comes from

Nakazaki E, Mah E, Sanoshy K, Citrolo D, Watanabe F (2021). Citicoline and Memory Function in Healthy Older Adults: A Randomized, Double-Blind, Placebo-Controlled Clinical Trial. Journal of Nutrition 151(8):2153–2160rct · Healthy adults 50–85 with age-associated memory impairment · checked 2026-08-23Citicoline (Cognizin) 500 mg/day for 12 weeks, 49 treated and 51 placebo. Episodic memory, assessed by the Paired Associate test, was a **secondary** outcome (0.15 against 0.06, p = 0.0025). NCT03369925. Cognizin is a Kyowa Hakko brand and two authors are affiliated with the sponsor.
Lee G, Choi S, Chang J, et al. (2021). Association of L-α Glycerylphosphorylcholine With Subsequent Stroke Risk After 10 Years. JAMA Network Open 4(11):e2136008cohort · Korean National Health Insurance members aged 50 and over without prior stroke or Alzheimer's disease · checked 2026-08-23After matching on all covariates, alpha-GPC users had adjusted hazard ratios of 1.43 (1.41–1.46) for total stroke, 1.34 for ischaemic and 1.37 for haemorrhagic stroke, rising with duration of use. Observational and confounded by indication; the authors call for work on the mechanism.
Poly C, Massaro JM, Seshadri S, et al. (2011). The relation of dietary choline to cognitive performance and white-matter hyperintensity in the Framingham Offspring Cohort. American Journal of Clinical Nutrition 94(6):1584–1591cohort · Framingham Offspring participants, dietary choline by food-frequency questionnaire · checked 2026-08-23An observational association between habitual dietary choline intake and cognitive performance. Association only; food-frequency intake is not a randomised dose.

And it costs nothing

The one thing worth doing

The cohort above was measuring food, not capsules — eggs, fish and soy are where habitual choline comes from. Eating them is the version of this page that costs nothing and is the only version with a population-sized dataset behind it.

Regulatory note — this note has not yet been reviewed by a lawyer. In the United States, supplements are sold under DSHEA: a structure/function claim is not reviewed by the FDA before it appears, and the label must carry the words "This statement has not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease." In the EU and the UK, only health claims on the authorised register may be used in advertising, and a food or supplement advertisement may not say a product prevents, treats or cures a disease. Whether an evidence page that carried a purchase link would itself count as advertising under those rules is an open question that counsel, not this page, will answer; there are no purchase links here. Nothing on this page is medical advice.