Ginkgo Biloba for Cognitive Impairment and Dementia: A Systematic Evidence Review

Does ginkgo biloba improve cognitive function in adults with or without cognitive impairment?

9 studies 7,471 participants 2008–2022

Evidence supports: Global Cognitive Function in Dementia/MCI, Alzheimer's Disease Cognitive Symptoms, Global Clinical Impression of Change in Dementia, Daily Function in Dementia +13 more

No clear effect: Attention and Concentration, General Memory Performance, Visuospatial Ability +4 more

Early data: Global Clinical Status in Dementia, Working Memory Capacity, Delayed Recall Memory

Abstract

Ginkgo’s clearest benefit is not dementia prevention and not broad cognitive enhancement. The evidence reviewed here shows that it can help manage established dementia symptoms, especially behavior, mood, sleep disruption, and some caregiver strain, and it may provide modest cognitive and functional support once impairment is already present.³⁴ Those benefits are real enough to matter, but they usually look more like smoother day to day functioning and less distress than a dramatic recovery of memory or thinking.

The strongest treatment signal is in neuropsychiatric symptoms. Across dementia trials, overall symptom burden improved by a moderate amount on average, including better apathy, fewer depressive symptoms, better sleep, and less disruptive behavior.³⁴ In one 24 week trial, 45% of people taking EGb 761 reached a clinically meaningful Neuropsychiatric Inventory improvement of at least 4 points, versus 23.8% on placebo, which means about 1 in 5 treated patients achieved that threshold because of treatment.³ Cognitive outcomes were less impressive. Global cognition improved only modestly, about a 1.4 point advantage on the MMSE, a 30 point screening test where 2 points is usually considered the minimum clearly meaningful change.⁸⁹ Daily function also improved slightly, while static global severity scales barely moved.⁴⁸

Outside established dementia, the picture gets narrower. Small trials suggest gains on some executive tasks such as set shifting and response inhibition, and one small study during ECT reported substantial short term memory benefits, but larger better powered trials in healthy older adults found no meaningful help for attention, broad memory, visuospatial ability, or overall cognitive decline.²⁵⁶⁷⁹

Most importantly, ginkgo does not appear to prevent dementia. The largest and longest trial, following 3,069 older adults for a median 6.1 years, found no reduction in all cause dementia, Alzheimer disease, or progression from mild cognitive impairment to dementia.¹² Taken together, ginkgo looks more like a symptom management option in established impairment than a prevention strategy.

In Plain Language

Ginkgo looks most useful for people who already have dementia symptoms, especially if the main problems are apathy, mood changes, sleep disruption, or difficult behavior.³⁴ It may also give a small boost in day to day function and thinking, but not a dramatic one.⁴⁸⁹

What it does not do, in the current analysis, is prevent dementia from happening in the first place.¹ It also does not reliably improve attention or general memory in otherwise healthy adults.²⁵

If someone is considering ginkgo, the most evidence-based reason is symptom support in established dementia, ideally using a standardized product similar to the 240 mg/day extracts studied in the trials, not for dementia prevention or as a general brain booster.³⁴

Introduction

The practical question is not whether ginkgo has any biological activity, it is whether that activity turns into noticeable cognitive or behavioral benefit in real people. That question matters because ginkgo is marketed both as a memory aid for healthy aging and as support for people already living with cognitive impairment, yet those are very different use cases.¹²⁴

The current analysis points to a split answer. Ginkgo shows its strongest and most consistent effects in people who already have dementia, where the main gains cluster in neuropsychiatric symptoms such as apathy, mood disturbance, sleep problems, and behavioral disruption.³⁴ By contrast, the prevention story is weak: the large long term trials in older adults do not show less dementia, less Alzheimer disease, or slower progression from mild cognitive impairment.¹²

That leaves a more nuanced middle ground. Ginkgo may modestly support cognition and daily function once impairment is present, but it does not look like a general cognitive booster in otherwise healthy adults. Smaller trials do show promising signals in selected tasks, especially executive control and some short term memory measures, but those sit inside a broader pattern of null results on attention, general memory, and global cognitive composites.²⁵⁶⁷⁹ The question is therefore not simply whether ginkgo works, but where it works, how much, and for what purpose.

Evidence 1 of 4

Behavioral relief and caregiver spillover in established dementia

Ginkgo shows its clearest value in dementia as a behavioral symptom treatment, not as a dramatic cognitive enhancer.³⁴ Across randomized trials, overall neuropsychiatric burden improved by a moderate amount, with a pooled effect that was statistically reliable and fairly consistent for this literature (standardized mean difference 0.46, 95% CI 0.32 to 0.61; I-squared 54.9%). I-squared measures how much study results differ from each other beyond chance, and here it means the benefit is not identical in every setting, but it points in the same general direction.³⁴ In the 24 week outpatient dementia trial, the Neuropsychiatric Inventory, or NPI, improved by 3.2 points with EGb 761 while placebo showed essentially no change, and 45.0% of treated patients achieved an NPI improvement of at least 4 points versus 23.8% on placebo, which means about 1 in 5 people gained a clearly meaningful behavioral benefit because of treatment.³

The most convincing symptom gains are in apathy, mood, sleep, and disruptive behavior.³ Apathy improved by a moderate to large amount, depressive symptoms improved by a moderate amount, and sleep disturbance improved by a similar moderate amount, with score changes large enough to cross typical meaningful-change thresholds in this analysis.³ Aberrant motor behavior showed an especially large signal, and irritability also improved, although that latter finding rests on less complete quantitative reporting.³ This pattern matters clinically because these symptoms often drive distress, supervision needs, and caregiver exhaustion more than memory test scores do.

Those behavioral gains appear to spill over into how families experience care, even if overall quality of life changes only a little.³⁴ Caregiver burden likely fell meaningfully in the available evidence, while patient quality of life improved only slightly, with a small effect on DEMQOL-proxy in the 2012 trial, rising by 3.3 points with ginkgo versus 1.1 with placebo over 24 weeks.⁴ That combination makes sense: reducing nighttime disturbance, apathy, and irritability can make daily care easier even when the person’s underlying disease remains clearly present.

The evidence is strongest when ginkgo is used in treatment-style dementia populations selected for behavioral symptoms, usually with standardized EGb 761 at 240 mg/day.³⁴ That does not guarantee the same result from every over-the-counter product, because extract standardization and dosing likely matter.

What this means

In established dementia, ginkgo most plausibly helps by making difficult days less difficult. The best-supported benefits are calmer behavior, less apathy, better sleep, and somewhat less caregiver strain, with only small gains in broad quality-of-life scores.³⁴

Dementia Neuropsychiatric Symptom Burden

Proven benefit Strong · 82

2 studies N=4,694 d_RE=0.46 (0.32 to 0.61) p=<.001 NNT=32.2 I²=55%

Proven but unnoticeable

R 2012 (n=333)

0.39

N 2011 (n=404)

0.52

Pooled

0.46

Favours control MCID Favours supplement

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	8 papers, majority low risk
Inconsistency	Serious	I²=55% (> 50%)
Imprecision	No concern	N=4694 meets OIS=400
Publication bias	No concern	k=6 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Dementia Quality of Life

Proven benefit Strong · 93

1 study N=1,139

Proven but unnoticeable

Single study: R 2012, d=0.29 (n=163+170)

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	2 papers, majority low risk
Inconsistency	No concern	no concerns (consistency=100%)
Imprecision	No concern	N=1139 meets OIS=400
Publication bias	No concern	k=2 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	High

Apathy

Proven benefit Strong · 94

0 studies N=1,598

Proven benefit

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	2 papers, majority low risk
Inconsistency	No concern	no concerns (consistency=100%)
Imprecision	No concern	N=1598 meets OIS=400
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	High

Depression or Dysphoria in Dementia

Proven benefit Strong · 94

0 studies N=1,598

Proven benefit

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	2 papers, majority low risk
Inconsistency	No concern	no concerns (consistency=100%)
Imprecision	No concern	N=1598 meets OIS=400
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	High

Dementia-Related Sleep Disturbance (CNPI Sleep Domain)

Proven benefit Strong · 94

0 studies N=1,598

Proven benefit

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	2 papers, majority low risk
Inconsistency	No concern	no concerns (consistency=100%)
Imprecision	No concern	N=1598 meets OIS=400
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	High

Aberrant Motor Behavior

Likely helps Strong · 70

0 studies N=331

Likely strong benefit

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	2 papers, majority low risk
Inconsistency	No concern	no concerns (consistency=100%)
Imprecision	Serious	N=331 below OIS=400
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Dementia Irritability/Lability

Likely helps Strong · 60

0 studies N=0

Likely helps, size unclear

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	1 papers, majority low risk
Inconsistency	No concern	single study, inconsistency N/A
Imprecision	Serious	sample size unknown
Publication bias	No concern	no d values
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Caregiver Burden in Dementia

Likely helps Strong · 71

0 studies N=1,628

Likely strong benefit

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	1 papers, majority low risk
Inconsistency	No concern	single study, inconsistency N/A
Imprecision	Serious	single study (N=1628), unreplicated
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Evidence 2 of 4

Modest cognitive and functional stabilization, not a stage reversal

Ginkgo suggests modest cognitive and functional benefit once impairment is present, but the gains are uneven and usually small.⁴⁸⁹ For global cognition in dementia or mild cognitive impairment, the average effect was positive, and the native-unit estimate translated to about a 1.4 point advantage on the MMSE, a 30 point cognitive screening test where 2 points is usually considered the minimum clearly meaningful change. That means the average change is probably real, but often subtle rather than restorative.⁸⁹ The pooled effect was statistically significant overall (0.54, 95% CI 0.17 to 0.91), but heterogeneity was extreme (I-squared 99.0%). In plain terms, some studies found more benefit than others, so the average effect should not be expected to show up equally strongly in every population.

Alzheimer-specific cognitive symptoms show a similar modest pattern, but with a major split between trials.²⁴ In the 24 week dementia treatment study, the SKT cognitive score improved by about 1.7 points relative to placebo in the Alzheimer subgroup, and 33% of treated patients achieved a clinically meaningful SKT response versus 14% on placebo.⁴ But the very large prevention-style older adult trial found no slowing of decline on ADAS-Cog over a median 6.1 years.² That is why the median signal looks favorable while the pooled confidence interval crosses no effect. A confidence interval is the range of effects compatible with the data, and when it crosses no effect, the average benefit is uncertain even if some individual trials are positive.

Clinician-rated overall improvement is more favorable than static severity scales, which suggests easing of symptoms and slower worsening rather than a clear shift in disease stage.⁴⁸ Global clinical improvement showed a positive signal across 4,632 participants, while global clinical status barely moved overall. This difference is not trivial. Improvement scales ask whether a person seems better or worse over time, whereas status scales capture where they sit on the severity ladder at a given moment. Ginkgo seems more likely to make someone look somewhat better than expected than to move them into a new severity category.

Daily independence likely improves slightly, but not by enough to transform self-sufficiency.⁴ In the 2012 dementia trial, ADL-IS, a scale of activities of daily living, improved by 0.16 points with ginkgo versus essentially no change with placebo over 24 weeks.⁴ The broader synthesis also favored treatment, and the prediction interval did not cross null, meaning future similar studies would still be expected to show benefit on average. Still, the effect size was small, so the likely experience is a bit more preserved day to day function, not a return to independent living.

The overall cognitive story in dementia is therefore supportive but restrained.²⁴⁸⁹ Ginkgo demonstrates some benefit for global improvement and suggests modest help for cognition and daily function, yet the evidence does not support dramatic reversal of impairment.

What this means

If ginkgo helps cognition in dementia, it usually looks like slower slipping or mild stabilization, not getting old abilities back. The average cognitive gain falls short of a clearly meaningful 2 point MMSE shift, but paired with behavioral improvement it may still be worthwhile for some patients.⁴⁸⁹

Global Cognitive Function in Dementia/MCI

Likely helps Good · 50

2 studies N=10,693 d_RE=0.54 (0.17 to 0.91) p=0.004 NNT=5.7 I²=99%

Likely modest benefit

M 2022 (n=80)

0.65

V 2017 (n=38)

0.32

Pooled

0.54

Favours control MCID Favours supplement

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	13 papers, majority low risk
Inconsistency	Serious	I²=99% (> 75%)
Imprecision	No concern	N=10693 meets OIS=400
Publication bias	Serious	Egger's p=0.000, funnel asymmetry detected (k=10)
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Low

Alzheimer's Disease Cognitive Symptoms

Likely helps Strong · 73

2 studies N=11,226 d_RE=0.18 (-0.28 to 0.65) p=0.432 I²=97%

Likely modest benefit

B 2009 (n=3069)

0.00

R 2012 (n=71)

0.49

Pooled

0.18

Favours control MCID Favours supplement

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	7 papers, majority low risk
Inconsistency	Serious	I²=97% (> 75%)
Imprecision	No concern	N=11226 meets OIS=400
Publication bias	No concern	k=6 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Global Clinical Impression of Change in Dementia

Proven benefit Strong · 78

0 studies N=4,632

Proven strong benefit

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	5 papers, majority low risk
Inconsistency	Serious	I²=96% (> 75%)
Imprecision	No concern	N=4632 meets OIS=400
Publication bias	No concern	k=2 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Global Clinical Status in Dementia

Not enough research Limited · 45

2 studies N=879 d_RE=0.19 (-0.06 to 0.44) p=0.127 I²=0%

Not enough research

D 2013 (n=210)

0.16

V 2017 (n=38)

0.36

Pooled

0.19

Favours control MCID Favours supplement

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	Serious	3/4 papers with RoB concerns
Inconsistency	No concern	no concerns (I²=0%, PI crosses null)
Imprecision	No concern	N=879 meets OIS=400
Publication bias	No concern	k=3 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Low

Daily Function in Dementia

Likely helps Good · 57

0 studies N=12,327 I²=84%

Likely modest benefit

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	10 papers, majority low risk
Inconsistency	Serious	I²=84% (> 75%)
Imprecision	No concern	N=12327 meets OIS=400
Publication bias	No concern	k=7 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Low

Evidence 3 of 4

Selective domain gains outside dementia, surrounded by nulls

Outside established dementia, ginkgo does not act like a general cognitive booster.²⁵⁶⁷⁹ The most reliable nulls are broad ones: attention and concentration showed no meaningful effect across 3,279 participants, with a pooled effect essentially at zero (0.02, 95% CI -0.05 to 0.09), and broad executive-function composites were also flat in the large older-adult trial (effect 0.01).²⁵ General memory performance was similarly unconvincing across 3,359 participants, and visuospatial ability was null in the 3,069 person prevention trial.²⁵⁹ These are the outcomes that matter most if someone is hoping for an across-the-board memory or focus upgrade, and they largely do not deliver.

The positive signals that do appear are narrower and more task-specific, especially in executive control.⁶⁷ Cognitive flexibility, meaning the ability to switch rules or mental sets, improved by a moderate amount overall. In one 8 week trial of adults aged 50 to 65 with subjective memory complaints, task-switching costs fell to 41 milliseconds with ginkgo versus 64 milliseconds with placebo, meaning participants lost less time when shifting between tasks.⁷ In a small acute crossover study of healthy young women, a single 120 to 240 mg dose reduced errors on Berg card sorting and Stroop tasks, which points to sharper set shifting and response control rather than a broad intelligence boost.⁶ Response inhibition also favored ginkgo overall, but the pooled effect was modest and the prediction interval crossed null, meaning the average benefit is plausible but may not repeat cleanly in every small study.⁵⁶⁷

Some of the largest apparent gains come from very small and unusual settings, which makes them promising rather than settled.⁹ During ECT treatment, 200 mg/day of ginkgo was associated with substantial short term improvements on WMS-III memory measures, including a 6.95 point gain from baseline versus 0.27 on placebo for total memory, and a 2.85 point MMSE improvement versus 1.27 on placebo.⁹ Those are large effects on paper, but they come from a single 80 person trial in a highly specific clinical context, with short follow-up and no long-term confirmation. The same study also showed that not every memory subscale improved, which is exactly what you would expect from an early signal rather than a settled broad effect.⁹

The mixed pattern is scientifically coherent. Executive tasks may be more sensitive to short term changes in cerebral blood flow, arousal, or frontally mediated control processes, while broad memory composites in large community samples are harder to move and more influenced by long-term neurodegeneration.²⁶⁷ Methodology probably contributes too. The larger trials tend to show smaller effects, while some of the strongest gains come from samples of 48 to 80 participants, where positive estimates are more likely to be exaggerated by chance.

The bottom line outside dementia is selective help, not general enhancement.²⁵⁶⁷⁹ Ginkgo may support certain executive or memory subdomains in particular contexts, but the current analysis does not support taking it for broad improvement in attention, general memory, or overall cognitive performance in otherwise healthy or earlier-stage adults.

What this means

Healthy adults or people with milder complaints should not expect ginkgo to sharpen everything. The best case is a narrow improvement in some executive tasks, with broad memory and attention outcomes mostly unchanged in the larger studies.²⁵⁶⁷

Executive Function

Proven benefit Strong · 94

1 study N=4,930

Proven but unnoticeable

Single study: B 2009, d=0.01 (n=1545+1524)

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	2 papers, majority low risk
Inconsistency	No concern	no concerns (consistency=100%)
Imprecision	No concern	N=4930 meets OIS=400
Publication bias	No concern	k=2 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	High

Cognitive Flexibility / Set Shifting

Proven benefit Strong · 90

3 studies N=418 d_RE=0.24 (-0.03 to 0.50) p=0.076 I²=6%

Proven benefit

S 2016 (n=61)

0.29

D 2016 (n=24)

0.76

D 2013 (n=140)

0.13

Pooled

0.24

Favours control MCID Favours supplement

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	4 papers, majority low risk
Inconsistency	No concern	no concerns (I²=6%, consistency=100%, PI crosses null)
Imprecision	No concern	N=418 meets OIS=400
Publication bias	No concern	k=4 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	High

Processing Speed

Likely helps Strong · 69

1 study N=273

Likely benefit

Single study: M 2022, d=0.74 (n=40+40)

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	2 papers, majority low risk
Inconsistency	No concern	no concerns (no data)
Imprecision	Serious	N=273 below OIS=400
Publication bias	No concern	k=2 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Immediate Memory and Learning

Likely helps Strong · 69

1 study N=273

Likely strong benefit

Single study: M 2022, d=1.50 (n=40+40)

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	2 papers, majority low risk
Inconsistency	No concern	no concerns (no data)
Imprecision	Serious	N=273 below OIS=400
Publication bias	No concern	k=2 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Response Inhibition

Likely helps Strong · 67

3 studies N=292 d_RE=0.27 (0.04 to 0.50) p=0.021 I²=0%

Likely modest benefit

S 2016 (n=58)

0.47

D 2013 (n=210)

0.19

D 2016 (n=24)

0.56

Pooled

0.27

Favours control MCID Favours supplement

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	3 papers, majority low risk
Inconsistency	No concern	no concerns (I²=0%, PI crosses null)
Imprecision	Serious	N=292 below OIS=400
Publication bias	No concern	k=3 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Working Memory Capacity

Early data Very early · 38

1 study N=80

Large effect, needs confirmation

Single study: M 2022, d=1.63 (n=40+40)

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	1 papers, majority low risk
Inconsistency	No concern	single study, inconsistency N/A
Imprecision	Very serious	single small study (N=80)
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Low

Delayed Recall Memory

Early data Limited · 40

0 studies N=193

Promising early signal

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	Serious	1/1 papers with RoB concerns
Inconsistency	No concern	single study, inconsistency N/A
Imprecision	Serious	single study (N=193), unreplicated
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Low

Attention and Concentration

No clear effect Strong · 94

2 studies N=3,279 d_RE=0.02 (-0.05 to 0.09) p=0.548

Doesn't appear to help

B 2009 (n=3069)

0.02

D 2013 (n=210)

0.10

Pooled

0.02

Favours control MCID Favours supplement

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	2 papers, majority low risk
Inconsistency	No concern	no concerns (no data)
Imprecision	No concern	N=3279 meets OIS=400
Publication bias	No concern	k=2 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	High

General Memory Performance

Likely no effect Strong · 71

3 studies N=3,359 d_RE=0.45 (-0.14 to 1.04) p=0.136 I²=90%

Probably doesn't help

B 2009 (n=3069)

0.01

D 2013 (n=210)

0.04

M 2022 (n=80)

1.47

Pooled

0.45

Favours control MCID Favours supplement

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	3 papers, majority low risk
Inconsistency	Serious	I²=90% (> 75%)
Imprecision	No concern	N=3359 meets OIS=400
Publication bias	No concern	k=3 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Visuospatial Ability

Likely no effect Strong · 72

1 study N=3,069

Probably doesn't help

Single study: B 2009, d=0.03 (n=1545+1524)

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	1 papers, majority low risk
Inconsistency	No concern	single study, inconsistency N/A
Imprecision	Serious	single study (N=3069), unreplicated
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Evidence 4 of 4

No prevention or disease-modification signal

Ginkgo does not appear to prevent dementia or stop mild cognitive impairment from worsening into dementia in the current analysis.¹² This is the most important reality check in the literature, because the prevention trials are the largest, longest, and least vulnerable to short-term placebo effects. In the 3,069 person randomized trial of adults aged 75 and older followed for a median 6.1 years, ginkgo did not reduce all-cause dementia (hazard ratio 1.12, 95% CI 0.94 to 1.33) or Alzheimer disease incidence (hazard ratio 1.16, 95% CI 0.97 to 1.39). A hazard ratio compares event rates over time, and values near 1.0 mean the two groups accumulated dementia at about the same pace.¹

The same study also found no meaningful slowing of progression among participants who already had mild cognitive impairment at baseline.¹ In that subgroup, progression to dementia was essentially unchanged (hazard ratio 1.13, 95% CI 0.85 to 1.50). Daily function in mild cognitive impairment was also effectively unchanged in the current synthesis.¹ Together, those findings argue against a disease-modifying effect.

A small vascular dementia signal in the prevention trial does not overturn the null overall conclusion.¹ Pure vascular dementia cases were fewer with ginkgo, but there were only 24 total cases, 7 versus 17, making the estimate too fragile to carry the main interpretation.¹ When the big primary outcomes are consistently null and the isolated positive finding rests on very few events, the honest read is that prevention has not been demonstrated.

This treatment-versus-prevention split is the central fact of the evidence base.¹²³⁴ Ginkgo may help people manage symptoms once dementia is present, but it does not show a convincing ability to keep dementia from starting or to halt its long-term trajectory.

What this means

Ginkgo should not be viewed as a dementia prevention strategy based on the current analysis. The best available long-term trial found no reduction in dementia onset, Alzheimer disease, or progression from mild cognitive impairment.¹

All-Cause Dementia Incidence

Likely no effect Strong · 72

1 study N=3,069

Probably doesn't help

Single study: S 2008, d=0.27 (n=1545+1524)

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	1 papers, majority low risk
Inconsistency	No concern	single study, inconsistency N/A
Imprecision	Serious	single study (N=3069), unreplicated
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Alzheimer Disease Incidence

Likely no effect Strong · 72

1 study N=3,069

Probably doesn't help

Single study: S 2008, d=0.08 (n=1545+1524)

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	1 papers, majority low risk
Inconsistency	No concern	single study, inconsistency N/A
Imprecision	Serious	single study (N=3069), unreplicated
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Progression from Mild Cognitive Impairment to Dementia

Likely no effect Strong · 68

1 study N=482

Probably doesn't help

Single study: S 2008, d=0.07 (n=256+226)

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	1 papers, majority low risk
Inconsistency	No concern	single study, inconsistency N/A
Imprecision	Serious	single study (N=482), unreplicated
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Daily Function in Mild Cognitive Impairment

Likely no effect Strong · 67

0 studies N=350

Probably doesn't help

▸ GRADE Assessment

Domain	Rating	Reason
Risk of bias	No concern	1 papers, majority low risk
Inconsistency	No concern	single study, inconsistency N/A
Imprecision	Serious	single study (N=350), unreplicated
Publication bias	No concern	k=1 usable (< 10), cannot assess per Cochrane 10.4
Indirectness	No concern	deferred to Phase 2 (#1546)
Overall certainty	Moderate

Across the Evidence

The evidence forms a clear pattern: ginkgo’s benefits cluster in symptom management after impairment has already developed, not in prevention before it appears.¹³⁴ That split makes biological sense. Behavioral symptoms such as apathy, sleep disruption, and irritability may be more responsive to short-term changes in neurotransmission, vascular function, or stress reactivity, while preventing dementia onset requires altering a long disease process that unfolds over many years. The treatment studies lasted about 24 weeks and often showed benefit; the prevention study lasted a median 6.1 years and still showed none.¹³⁴

The outcome pattern is also domain-specific rather than global.²³⁴⁶⁷⁹ Ginkgo helps some things much more than others. In dementia, behavior and mood improve most consistently. In non-dementia settings, isolated executive-control tasks sometimes improve while broad attention, memory, and visuospatial composites do not. Different instruments are detecting different layers of change. A clinician or caregiver may notice a calmer, more engaged, better-sleeping person before a broad cognitive battery shows much movement.

The most impressive numbers often come from the smallest studies, which lowers confidence that the apparent size will hold up in routine practice.⁶⁷⁹ Several pooled outcomes had very high heterogeneity, including global cognition in dementia or MCI (I-squared 99.0%), Alzheimer cognitive symptoms (96.9%), and general memory performance (90.1%). Heterogeneity means studies are not all estimating the same practical effect. Some of that is population mixing, some is measurement choice, and some may reflect formulation and dose differences. It tells us that averaging the trials is informative, but it does not justify expecting one uniform response.

Dose and formulation probably matter.¹³⁴⁷⁸ The most practically useful positive literature is concentrated around standardized EGb 761, especially 240 mg/day, in established dementia. That matters because supplement markets are messy. A favorable result with a standardized extract used in clinical trials should not be assumed to apply to every product labeled ginkgo.

There is also a consistent difference between statistical reliability and felt impact.³⁴⁸⁹ Some outcomes are clearly positive in the data but still fall below typical thresholds for an obvious day-to-day transformation. That is why the overall picture is best described as easing symptoms, smoothing function, or modestly slowing deterioration, rather than restoring normal cognition.

Discussion

The evidence reviewed here supports ginkgo as a selective symptom-management option, not a broad cognitive enhancer and not a dementia prevention tool.¹²³⁴ Confidence is highest for behavioral symptoms in established dementia, where the effects are consistent enough and large enough to matter, especially for apathy, depressive symptoms, sleep disturbance, and overall neuropsychiatric burden.³⁴ Confidence is moderate for modest cognitive and functional benefit in already impaired populations. That benefit appears real on average, but it is small and variable across studies, and it does not amount to reversal of disease severity.²⁴⁸⁹

The prevention question is much closer to settled, and the answer is negative in this body of evidence.¹² A large, well-run long-term randomized trial found no reduction in all-cause dementia, Alzheimer disease, or progression from mild cognitive impairment, and that matters more than scattered smaller positive signals on isolated cognitive tasks.¹ If future studies are going to change this conclusion, they will need to be large, long, and clearly better than the existing prevention evidence, not merely another short-term memory test study.

The biggest uncertainties are outside dementia and in niche populations.⁵⁶⁷⁹ Small trials hint at benefits in cognitive flexibility, response inhibition, short-term learning, and ECT-related cognitive outcomes, but these findings need replication because they often come from 48 to 80 participants and from highly specific contexts. Early positive estimates are common in supplement research and often shrink in larger confirmatory trials.

GRADE rated several outcomes as low or moderate certainty, and that can sound more discouraging than the full picture really is. GRADE is appropriately strict, but it tends to punish nutrition and supplement evidence because it heavily downgrades observational evidence and does not give much credit unless effects are very large. Here, that means some outcomes can look clinically useful even when GRADE still says low certainty. The right interpretation is not that the findings are meaningless, but that they are promising or useful with limits, especially when study results vary a lot or replication is thin.

If stronger evidence emerges, the most likely place it will do so is not in prevention but in more precise dementia treatment questions: which patients benefit most, whether behavioral symptoms predict response, and whether standardized 240 mg/day extracts outperform lower-dose or less standardized products.³⁴⁷⁸ For now, what is supported is narrower but still meaningful: ginkgo may help some people with established dementia function a little more smoothly and with less behavioral distress, but it is not a substitute for disease-modifying therapy and should not be expected to prevent decline from starting.

Methodology

We searched PubMed for studies on ginkgo and cognitive impairment or dementia, then filtered to controlled human studies that matched the review question. Nine randomized trials were included from a larger screened literature. We read each study, recorded who was studied, what dose and formulation were used, what outcomes were measured, how large the trials were, and what they found. We assessed evidence quality with the GRADE framework and judged clinical importance against published meaningful-change thresholds when those were available.

GRADE was built mainly for pharmaceutical interventions, and it systematically rates nutrition and supplement evidence conservatively. It automatically downgrades observational research and usually only upgrades for very large effects, often above a relative risk of 2.0. Because of that, supplement evidence can be labeled low certainty even when the practical signal is fairly consistent. Our trust score adds a continuous view of how persuasive and clinically meaningful the evidence looks, including whether the average effect reaches thresholds people are likely to notice. When GRADE is low but the trust signal is stronger, that usually reflects real but modest benefits, outcome-specific consistency, or clinically relevant change that does not fit GRADE’s upgrade rules.

Every study cited here is publicly indexed on PubMed. Known limitations include mixed populations, different cognitive tests that are not directly interchangeable, high heterogeneity for several pooled outcomes, and many non-dementia findings coming from small single trials.

Study Selection

130 Papers in ginkgo corpus

13 wrong study type, 2 wrong comparator

24 With matching outcomes

9 After study type & comparator filters

9 Included in review

Characteristics of Included Studies

Study	Design	N	Population	Dose	Duration	RoB
S 2008 FT	rct	3069	subclinical	240 mg daily (120 mg twice daily)	median follow-up 6.1 years (maximum 7.3 years)	Low
B 2009 FT	rct	3069	subclinical	120 mg twice daily	Median follow-up 6.1 years (maximum 7.3 years)	Low
N 2011 FT	rct	404	clinical	240 mg once daily for 24 weeks	24 weeks	Some
R 2012 FT	rct	410	clinical	240 mg once daily for 24 weeks	24 weeks	Some
D 2013 FT	rct	226	clinical	120 mg/day (60 mg twice daily)	24 months follow-up	Some
D 2016 FT	rct	48	healthy	120 mg, single dose	acute single-dose sessions on three days (each testing day separated by 48 h washout); measurements taken up to ~60+ minutes post-dose during cognitive testing	Low
S 2016 FT	rct	75	healthy	240 mg daily for 8 weeks	56 days (56 )	Low
V 2017 FT	rct	90	clinical	120 mg daily for 6 months	6 months	High
M 2022 FT	rct	80	clinical	200 mg daily during ECT	6–12 ECT sessions (2–3 sessions per week); Ginkgo given from 48 hours before first ECT until end of ECT course	Some

Sources