We ranked memory supplements by randomized trial evidence. The winner is boring: a daily multivitamin. The loser is famous: ginkgo biloba. Here are the numbers from 6,750 participants.
In plain terms A cheap daily multivitamin showed small but real memory benefits in large trials. Ginkgo biloba did nothing in a six-year study of 3,069 people. B vitamins help if your homocysteine is high.
COSMOS-Mind enrolled 2,262 adults aged 65 and older and followed them for three years[1]. Participants took either a daily multivitamin or an identical placebo. A parallel arm tested cocoa flavanols. Cognitive tests were administered by phone annually. The trial was double-blind and placebo-controlled with preregistered endpoints.
The multivitamin group improved global cognition by 0.07 standard deviations compared to placebo[1]. The study authors equated this effect size to offsetting roughly 1.8 years of typical age-related cognitive decline. The cocoa extract arm showed no benefit on any cognitive measure.
A companion trial, COSMOS-Clinic, tested 573 participants with in-person neuropsychological batteries rather than phone assessments[2]. Results confirmed the multivitamin benefit with more rigorous measurement.
Pooled analyses across COSMOS trials included 5,203 participants total[2]. Global cognition improved by 0.07 standard deviations. Episodic memory specifically improved by 0.06 standard deviations.
The effect size is modest in absolute terms. A 0.07 SD change would be undetectable to an individual on any given day. At the population level across years of aging, small protective effects compound. The comparison is to slowing decline, not to a cognitive enhancer.
Practical context matters. The tested product was a standard daily multivitamin with RDA-level doses. Cost runs a few cents per day. Risks are minimal for most adults.
Caveats remain. Participants were generally well-educated volunteers. Effects in populations with poor baseline nutrition might differ. Which nutrient drives the benefit, or whether the combination matters, is unknown.
Swipe the chart sideways
| Intervention | Study | Participants | Duration | Result |
|---|---|---|---|---|
| Multivitamin | COSMOS-Mind + COSMOS-Clinic, RCT | 5,203 pooled | 3 years | Global cognition +0.07 SD, episodic memory +0.06 SD[1][2] |
| Ginkgo biloba | GEM, RCT | 3,069 | 6.1 years | No reduction in dementia or Alzheimer's incidence[3] |
| Creatine | Meta-analysis of 16 RCTs | 492 | Days to weeks | Memory SMD 0.31; overall cognition and executive function null[5] |
| B vitamins | VITACOG, RCT in MCI | 271 | 2 years | Brain atrophy 1.08% to 0.76% per year; 53% slower if homocysteine above 13 umol/L[4] |
| MCT ketogenic drink | RCT, crossover, in MCI | 83 | 6 months | Improved recall, fluency, naming, trail-making[6] |
All six sources were verified against PubMed on September 6, 2026.
The GEM trial remains the largest and longest randomized test of ginkgo for dementia prevention[3]. It enrolled 3,069 adults aged 75 and older. Participants received either 240 mg of standardized ginkgo extract daily or placebo. Follow-up averaged 6.1 years with dementia diagnosis as the primary endpoint.
The trial was rigorous. Dementia outcomes were adjudicated by expert panels blinded to treatment assignment. The sample size was powered to detect clinically meaningful reductions in dementia incidence.
Ginkgo did not reduce the rate of dementia or Alzheimer's disease compared to placebo[3]. The hazard ratio crossed one with no trend toward benefit. Secondary cognitive outcomes also showed no advantage.
This null result contradicted decades of smaller trials and observational studies suggesting ginkgo might protect cognition. The GEM trial's size and duration made it the definitive test. Earlier positive findings likely reflected publication bias or confounding.
For buyers, this evidence is difficult to dismiss. A 6-year trial with over 3,000 participants showing no effect is stronger than multiple small positive studies.
Ginkgo is generally safe but carries bleeding risks, particularly with anticoagulants. Given the null efficacy data, even modest risks shift the benefit-harm balance unfavorably.
VITACOG randomized 271 adults with mild cognitive impairment to high-dose B vitamins or placebo for two years[4]. The active treatment contained 0.8 mg folic acid, 0.5 mg vitamin B12, and 20 mg vitamin B6.
The primary outcome was brain atrophy measured by MRI. The placebo group lost brain volume at 1.08 percent per year[4]. The B vitamin group lost volume at 0.76 percent per year. That is a 30 percent relative reduction in the rate of brain shrinkage.
The benefit depended on baseline homocysteine levels. Participants with homocysteine above 13 micromoles per liter experienced 53 percent slower atrophy. Those below 13 showed no significant benefit.
This interaction is biologically coherent. B vitamins lower homocysteine through the methylation cycle. If homocysteine is already normal, there is nothing to correct.
Practical implications are clear. Test homocysteine before supplementing. A simple blood test costs relatively little. If levels are elevated, B vitamins may help. If normal, they likely offer no cognitive protection.
The trial population had mild cognitive impairment, not normal cognition or dementia. Brain atrophy is a surrogate outcome. Whether slowing atrophy preserves daily function remains uncertain.
B vitamins at these doses are safe for most people. Very high B6 intake over years can cause peripheral neuropathy; the VITACOG doses stayed within ranges used in other trials without reported nerve damage.
Swipe the chart sideways
A 2024 meta-analysis pooled 16 randomized controlled trials testing creatine for cognition[5]. The combined sample included 492 adults. Most trials used doses around 5 g per day.
Memory outcomes improved with a standardized mean difference of 0.31[5]. That is a small to moderate effect. Attention and processing speed also improved.
Overall global cognition scores did not change significantly. Executive function measures were also null. Creatine appears to affect specific cognitive domains rather than broad intellectual capacity.
The likely mechanism is brain energy metabolism. Creatine helps regenerate ATP in cells with high energy demand. Neurons during demanding tasks may benefit from enhanced phosphocreatine reserves.
Creatine monohydrate is the most studied form. Five grams daily is standard. Loading phases used in athletic contexts are probably unnecessary for cognitive purposes.
Creatine is among the safest supplements studied. Decades of use in athletes show minimal side effects in healthy people.
Caveats apply. The meta-analysis pooled heterogeneous populations and protocols, from young adults to older adults, single doses to weeks of use. Effect sizes in any specific subgroup may differ.
A six-month trial tested a ketogenic medium-chain triglyceride drink in 83 adults with mild cognitive impairment[6]. Participants consumed 30 g of MCT oil daily or a calorie-matched placebo. The goal was to raise blood ketones as an alternative brain fuel.
The MCT group improved on several cognitive tests compared to placebo[6]. Recall performance increased. Verbal fluency scores rose. Naming tasks showed gains. Trail-making times improved.
The rationale involves brain energy metabolism. In aging and Alzheimer's pathology, glucose uptake in the brain declines. Ketones can bypass this deficit without full dietary ketosis.
Thirty grams daily is a meaningful dose. Lower doses in some commercial products may not reach the ketone levels tested.
Gastrointestinal side effects are common with MCT oil: nausea, cramping, diarrhea. Gradual escalation and taking it with food help.
This trial studied people with existing mild cognitive impairment, not healthy adults. Benefits in other populations are unproven. Six months is short for a neurodegenerative condition. Replication in larger samples would strengthen confidence.
Pooled N: 5,203 adults
Effect: global cognition +0.07 SD
Largest evidence base of any memory supplement. Costs cents per day. Effects are small but replicated across two independent cohorts.
Evidence tier BRead the COSMOS-Clinic trialTrial N: 3,069 adults
Effect: no benefit vs placebo
The largest and longest test of the most famous memory herb found nothing. Bleeding risk with anticoagulants adds harm without benefit.
Evidence tier DRead the GEM trialTrial N: 271 adults with MCI
Effect: 53% slower atrophy if Hcy above 13 umol/L
Benefit is conditional. Get a homocysteine blood test before supplementing. Normal levels means no measured benefit.
Evidence tier BRead the VITACOG trialPooled N: 492 adults, 16 RCTs
Effect: memory SMD 0.31
Improves memory and processing speed. Overall cognition and executive function stayed null. 5 g/day monohydrate is the studied form.
Evidence tier BRead the meta-analysisNo trial in this set has shown dementia prevention. All measured effects are small. The practical floor is free: exercise, sleep, and blood pressure control have larger effect sizes than anything sold in a bottle.
Global cognition improved by 0.07 standard deviations over 3 years in COSMOS-Mind, with 2,262 participants. The authors compared the size to offsetting roughly 1.8 years of typical age-related decline. The effect is real but small, and it replicated in a second cohort of 573 adults tested in person.
The GEM trial followed 3,069 adults aged 75 and older for a median of 6.1 years on 240 mg/day of standardized extract. Dementia and Alzheimer's incidence did not differ from placebo. Popularity is not an endpoint.
Adults with elevated homocysteine. In VITACOG, 271 adults with mild cognitive impairment took folic acid 0.8 mg, B12 0.5 mg and B6 20 mg for 2 years. Brain atrophy slowed from 1.08 to 0.76 percent per year. With homocysteine below 13 umol/L, there was no significant benefit.
No. A meta-analysis of 16 trials with 492 adults found memory improved with an effect size of 0.31, and attention and processing time improved. Overall cognition and executive function showed no significant effect.
Corrections and source queries: [email protected].