Sudoku and Brain Health: What the Research Actually Says

Almost every article about puzzles and brain health repeats the same three claims: puzzles make you smarter, keep your brain 10 years younger, and prevent dementia. Most of those articles cite each other rather than the underlying studies. This page goes back to the primary sources instead, states what each study actually found, and is just as clear about what the evidence does not show. It is maintained as a living reference: every claim links to the original paper, every source is re-verified on each review, and the update log at the bottom records what changed. Feel free to cite it.

Executive summary

• In a 19,078-person observational study, older adults who did number puzzles regularly scored better on cognitive tests than those who never did. This is a correlation; causation has not been shown. Rating: Moderate for the association, not for cause.
• The famous “brain function 10 years younger” line comes from a university press release, not from the peer-reviewed papers it describes.
• Deliberate cognitive training produces real, durable gains in the trained skills. A 2,832-person randomized trial found effects still present ten years later. Rating: Strong, for trained skills specifically.
• One randomized trial found ordinary crossword puzzles beat commercial brain-training games in adults with mild cognitive impairment. Rating: Moderate (a single trial, in an impaired population, using word puzzles).
• Broad transfer from puzzles to general intelligence or everyday cognition remains unproven. The most thorough review of the field found little convincing evidence for it. Rating: No evidence of transfer.
• No study has shown that sudoku prevents dementia, and the Lancet Commission does not list puzzles among its fourteen modifiable risk factors. Rating: No evidence.
• The first randomized trial of sudoku itself (SMART, University of Hong Kong, 262 participants) is completed but unpublished. This page will be updated when its results appear.

What’s new since the last review

Checked 23 July 2026. All ten sources below were re-verified against their registries and publishers on this date. The SMART sudoku trial (NCT04913857) remains completed but unpublished; a search of the trial registry and PubMed found no posted results and no journal publication. No new randomized trials of sudoku or number puzzles were found since the previous review. Nothing in the conclusions has changed.

How we rate the evidence

Each claim on this page carries one of four plain-language labels, loosely modeled on the GRADE approach used in medical guidelines:

• Strong: at least one large randomized trial with long follow-up, pointing the same way as the rest of the literature.
• Moderate: consistent findings from large observational studies, or a single well-run randomized trial that has not yet been replicated.
• Limited: small studies, correlational designs, or findings that have not been independently repeated.
• No evidence: the claim circulates widely but no study supports it, or the best available review points the other way.

A rating describes the state of the research, not our opinion of sudoku. We run a sudoku site; the ratings below are deliberately conservative for exactly that reason.

The evidence at a glance

Myth vs evidence

Timeline of the research

• 2011: Grabbe reports a correlation between sudoku performance and working memory in older adults. The first study to look at sudoku specifically.
• 2014: The ACTIVE trial publishes 10-year follow-up results: cognitive training effects in older adults persist a decade.
• 2016: Simons and six colleagues publish the field’s most thorough review and find little evidence that brain training transfers beyond the trained task.
• 2019: The PROTECT papers link puzzle frequency to cognitive scores in 19,078 adults. The “10 years younger” press release travels around the world.
• 2020 to 2023: The University of Hong Kong runs SMART, the first randomized trial of a structured sudoku program, in 262 adults with mild cognitive impairment.
• 2022: NEJM Evidence publishes the crosswords-versus-games trial: the plain puzzle wins.
• 2024: The Lancet Commission updates its dementia-prevention report: fourteen risk factors, puzzles not among them.
• Now: SMART results awaited. When they publish, this page gets its first major update.

The evidence in detail

The big study everyone quotes, and what it really found

In 2019, researchers from the University of Exeter and King’s College London published two companion papers in the International Journal of Geriatric Psychiatry, drawing on 19,078 adults aged 50 to 93 in the online PROTECT cohort. One paper looked at number puzzles such as sudoku, the other at word puzzles. Both found the same pattern: the more often people reported doing puzzles, the better they scored across fourteen measures of attention, reasoning and memory. People who never did puzzles scored worst.

That is a genuinely large and interesting result, and it deserves two honest caveats. First, it is cross-sectional: puzzle habits and test scores were measured at the same time, so the data cannot tell whether puzzles sharpen minds or sharp minds enjoy puzzles. Independent experts made exactly this point when the papers came out, as collected by the UK Science Media Centre. Second, the famous “ten years younger” figure appears in the university’s press release, where the lead researcher translated the score gaps into age equivalents. It is a fair illustration, but it is not a finding stated in the papers, and note that both papers describe the same 19,078 people, so they are one study seen from two angles, not two independent confirmations.

What randomized trials show

Correlations cannot prove cause. Randomized trials can, and there are two worth knowing.

The ACTIVE trial randomized 2,832 older adults to structured training in memory, reasoning or processing speed, or to no training. Ten years later, the reasoning and speed groups still performed better on the specific abilities they had trained, and all trained groups reported less decline in daily-living activities. The training was a formal curriculum rather than puzzles, but it is the strongest evidence that deliberate mental practice in later life produces durable, measurable gains.

More directly relevant, a 2022 trial in NEJM Evidence randomized 107 older adults with mild cognitive impairment to either commercial computerized brain-training games or ordinary crossword puzzles for 78 weeks. The crossword group did better: their cognitive scores improved slightly while the games group declined slightly, and the difference reached statistical significance. Two cautions apply. The participants already had mild impairment, so the result may not generalize to healthy adults, and the trial tested word puzzles, not sudoku. Still, it is striking that a plain, cheap puzzle beat purpose-built training software.

The skeptics’ case, taken seriously

In 2016, seven researchers published the most thorough review of the brain-training literature, in Psychological Science in the Public Interest. Their conclusion is the one every puzzle site should quote and almost none does: training reliably improves performance on the trained task, sometimes improves closely related tasks, and shows little convincing evidence of transfer to general intelligence or everyday cognitive life. Practicing sudoku makes you better at sudoku, and at the reasoning patterns sudoku exercises. It has not been shown to make you better at remembering names or balancing a budget. Our shorter overview, Does Sudoku Make You Smarter?, walks through what this means for everyday players.

Research on sudoku itself

Surprisingly little research studies sudoku specifically. A small 2011 study by Grabbe in Activities, Adaptation and Aging reported that sudoku performance correlated with working-memory measures in older adults, which suggests the puzzle exercises those systems but again cannot show cause. More promising, the University of Hong Kong has completed a proper randomized trial, the SMART study, which gave 262 adults aged 55 and over with mild cognitive impairment a 24-week structured sudoku training program and measured cognition against a wait-list control group. The trial finished at the end of 2023 and results have not yet been posted. When they are published, this will be the first direct causal evidence about sudoku either way, and we will update this page accordingly.

What about dementia prevention?

The authority here is the Lancet Commission on dementia prevention, whose 2024 report estimates that around 45 percent of dementia cases are linked to fourteen modifiable risk factors. Puzzles are not one of them. The list is things like hearing loss, hypertension, smoking, physical inactivity, social isolation and low education. A cognitively active life fits the spirit of those recommendations, and puzzles are a pleasant part of such a life, but any site telling you sudoku prevents dementia is claiming something no study has shown. That includes ours: we make no such claim.

Our honest bottom line

Regular puzzlers test better than non-puzzlers on a large scale, deliberate mental practice produces real gains in the skills practiced, one trial found ordinary puzzles beat commercial brain training in impaired adults, and broad transfer to general intelligence remains unproven. Sudoku is a workout for working memory, attention and logical reasoning, it is free, and it is enjoyable, which is the reason to do it. You can play it free in your browser, make it a habit with the daily challenge, or print a batch for offline solving. Older solvers may prefer the large-print seniors mode. If the SMART trial shows more than that, you will read it here first.

How to cite this page

Writers and researchers are welcome to quote or link to this page. Suggested citation: “Sudoku and Brain Health: What the Research Actually Says.” SudokuPuzzleHub, last reviewed July 2026. https://sudokupuzzlehub.com/blog/sudoku-brain-health-research/. If you quote a specific finding, please cite the primary study alongside; every entry in the research database below links to its original source.

Research database

1. Brooker H, et al. Number-puzzle use and baseline cognitive function in adults 50+. Int J Geriatr Psychiatry. 2019;34(7):932-940. Cross-sectional, n=19,078. https://doi.org/10.1002/gps.5085
2. Brooker H, et al. Word puzzle use and cognitive function in older adults. Int J Geriatr Psychiatry. 2019;34(7):921-931. Cross-sectional, same cohort as [1]. https://doi.org/10.1002/gps.5033
3. University of Exeter press release, 16 May 2019. Source of the “10 years younger” quote. news-archive.exeter.ac.uk
4. Science Media Centre expert reaction, 16 May 2019. Independent commentary on [1] and [2]. sciencemediacentre.org
5. Rebok GW, et al. Ten-year effects of the ACTIVE cognitive training trial. J Am Geriatr Soc. 2014;62(1):16-24. RCT, n=2,832. https://doi.org/10.1111/jgs.12607
6. Devanand DP, et al. Computerized games versus crosswords training in MCI. NEJM Evidence. 2022;1(12). RCT, n=107, 78 weeks. https://doi.org/10.1056/EVIDoa2200121
7. Simons DJ, et al. Do “brain-training” programs work? Psychol Sci Public Interest. 2016;17(3):103-186. Systematic review. https://doi.org/10.1177/1529100616661983
8. Grabbe JW. Sudoku and working memory performance for older adults. Act Adapt Aging. 2011;35(3):241-254. Correlational; sample size unverified (paywalled), which is why this page cites no number for it. https://doi.org/10.1080/01924788.2011.596748
9. SMART trial, NCT04913857, University of Hong Kong. RCT, n=262, completed December 2023, results pending. clinicaltrials.gov/study/NCT04913857
10. Livingston G, et al. Dementia prevention: 2024 Lancet Commission report. Lancet. 2024;404(10452):572-628. https://doi.org/10.1016/S0140-6736(24)01296-0

Update log

This page is reviewed when new primary research is published, and at least twice a year. Maintained by the team behind SudokuPuzzleHub’s free puzzles; read more about us. Corrections are welcome via the community.