Your parents do a crossword every morning. They say it keeps their mind sharp. You nodded and assumed it was a comforting story.
It is not just a story anymore. The clinical trials now say they are onto something, with one important twist: the type of puzzle matters, and crosswords specifically outperformed the computerized brain games they were tested against.
The Short Answer
Yes, word games can slow measurable cognitive decline in people with mild cognitive impairment, and the strongest 2026 evidence points to crosswords specifically. A randomized trial published in NEJM Evidence found that home-based crossword training improved cognition and daily functioning in adults with mild cognitive impairment over 78 weeks, while computerized cognitive games did not. A follow-up trial, COGIT-2, is now testing the optimal dose. Word games are not a cure, and they do not prevent dementia in healthy adults, but for people already showing mild impairment, the evidence is real.
The Study That Changed the Conversation
The main trial appeared in NEJM Evidence. Researchers at two sites enrolled 107 adults with mild cognitive impairment. MCI is the stage between normal aging and dementia, where memory and thinking are below expected but daily function is largely preserved. It raises the risk of progressing to dementia.
Participants were randomized to either web-based cognitive games or web-based crossword puzzles, delivered through the CogniFit platform. The trial ran for 78 weeks, with 12 weeks of intensive training followed by booster sessions.
The primary outcome was change on the ADAS-Cog, the 11-item Alzheimer's Disease Assessment Scale, where higher scores mean greater impairment.
The result: the crossword group improved. The games group worsened slightly. The difference was statistically significant. The crossword group also showed better daily functioning and less brain atrophy on MRI scans, with greater hippocampal volume and cortical thickness preserved compared to the games group.
This was not a small survey. It was a randomized controlled trial published in a peer-reviewed journal, and it directly compared two active interventions.
Why Crosswords Beat Computerized Games
The finding surprised people who assumed modern, gamified brain training would outperform an old-fashioned puzzle. The researchers offered a few reasons.
Crosswords are language-based and tap into crystallized intelligence, the accumulated knowledge and vocabulary that tends to hold up longer in aging brains. Computerized cognitive games often train fluid intelligence, processing speed and novel problem-solving, which declines earlier and is harder to retrain.
Crosswords also have a difficulty calibration that worked well. The trial used puzzles of moderate difficulty, challenging enough to engage the solver without causing frustration that drives dropout. The structure of a crossword, with clues, crossings, and a completable grid, may also provide a sense of mastery that abstract reaction-time tasks lack.
The COGIT-2 Follow-Up Trial
A second trial, COGIT-2, registered as NCT06601933 and published in the International Journal of Clinical Trials in April 2025, is now extending the finding. It enrolls 240 adults with MCI across multiple sites and compares three groups over 78 weeks.
- High-dose crosswords: four puzzles per week
- Low-dose crosswords: one puzzle per week
- Health education control
The primary outcome is change on ADAS-Cog14. Secondary outcomes include daily functioning, MRI hippocampal volume, cortical thickness, and blood biomarkers including plasma pTau217, a marker tied to Alzheimer's pathology.
COGIT-2 is designed to answer the dose question: how much crossword practice is enough? If four puzzles per week beat one, the recommendation becomes specific. The trial is also the first to pair behavioral outcomes with imaging and biomarker data, which would show whether crossword practice changes the brain's physical trajectory, not just test scores.
The 2026 Meta-Analysis on Video Games and Cognition
A separate meta-analysis published in Frontiers in Aging Neuroscience in January 2026 reviewed five randomized controlled trials of video-game interventions in adults with MCI, covering about 215 participants. It found significant improvements in global cognition, measured by MoCA and MMSE scores, plus gains in executive function and attention.
The effect sizes were moderate to large. But the authors stressed a caveat: the number of studies was small, and the interventions varied widely. The meta-analysis supports the broader claim that structured cognitive engagement helps, while the crossword-specific trial shows that not all cognitive engagement is equal.
What the Research Does Not Claim
This matters as much as what it does claim.
Word games do not prevent dementia in healthy adults. The strongest findings are in people who already have mild cognitive impairment. There is no trial showing that daily crosswords stop healthy adults from ever developing dementia.
Word games are not a treatment on their own. They are a low-cost, low-risk adjunct. The researchers frame them as a scalable, home-based tool, not a replacement for medical care.
Benefits are specific to the trained domain. Doing crosswords makes you better at crosswords and improves related language and memory measures. It does not broadly upgrade every cognitive function. Transfer is limited, which is consistent with decades of brain-training research.
More research is still coming. COGIT-2 has not reported final outcomes. The video-game meta-analysis covered only five trials. The picture is encouraging but incomplete.
How This Fits the Earlier Evidence
A 2019 study in the International Journal of Geriatric Psychiatry analyzed over 17,000 adults aged 50 and older. Frequent word-puzzle solvers performed better on attention, reasoning, and memory tests, scoring as if they were roughly ten years younger cognitively than non-solvers. That was observational, so it could not prove causation. Solvers might simply be people who started with sharper minds.
The 2026 NEJM Evidence trial addresses that gap with randomization. By assigning people to crosswords or games by chance, it rules out the self-selection explanation. The crossword group's improvement is attributable to the crosswords, not to preexisting differences.
Practical Takeaways for Players
If you or someone you know is concerned about cognitive decline, the research suggests a few concrete steps.
- Do word puzzles regularly, not occasionally. The COGIT-2 dose comparison suggests frequency matters. Four sessions a week is being tested against one. More is likely better, up to a point.
- Choose puzzles with moderate difficulty. Too easy and you disengage. Too hard and you quit. Crosswords calibrate naturally because you fill in what you can and let crossings help with the rest.
- Prefer language-based puzzles for this specific goal. The trial favored crosswords over abstract cognitive games. That does not mean other games are useless, only that the strongest evidence is for word puzzles.
- Pair puzzles with other healthy behaviors. Exercise, sleep, social engagement, and blood pressure control all have their own evidence base for brain health. Puzzles are one layer, not the whole stack.
Comparison: What Different Puzzles Offer
| Puzzle type | Cognitive demand | Evidence for slowing decline |
|---|---|---|
| Crosswords | Language, memory, recall | Strongest, from RCT |
| Cryptograms | Pattern recognition, deduction | Indirect, via frequency analysis |
| Word-guessing (Lexle) | Deduction, vocabulary | Indirect |
| Spelling Bee | Vocabulary, search | Indirect |
| Speed games (Typing Rush) | Processing speed, motor | Mixed in trials |
The pattern is that language-rich puzzles carry the strongest evidence, while speed and reaction tasks show smaller or more inconsistent effects.
FAQ
Can word games prevent dementia?
No study shows that word games prevent dementia in healthy adults. The strongest evidence is for slowing measurable decline in people who already have mild cognitive impairment.
Which word game has the most evidence behind it?
Crosswords. The NEJM Evidence randomized trial specifically tested crosswords against computerized cognitive games and found crosswords superior for people with MCI.
How often should you do word puzzles for brain health?
The COGIT-2 trial is testing four puzzles per week against one. More frequent practice appears more beneficial, though the optimal dose is still being established.
Do brain-training apps work as well as crosswords?
For people with mild cognitive impairment, the trial evidence says no. Crosswords outperformed computerized cognitive games on cognition, daily function, and brain atrophy measures over 78 weeks.
Are word games a substitute for medical treatment?
No. They are a low-cost, low-risk adjunct. Anyone concerned about memory loss should consult a clinician, not rely on puzzles alone.
The Takeaway
Word games can slow measurable cognitive decline in people with mild cognitive impairment, and the 2026 evidence points to crosswords specifically. A randomized trial in NEJM Evidence found crosswords improved cognition and preserved brain volume better than computerized games over 78 weeks. A follow-up trial, COGIT-2, is now pinning down the right dose. Word games are not a cure or a guarantee, but for people at risk, the case for regular language-based puzzles is now backed by real clinical data.
Start a daily habit on the Crossword or Cryptogram, and rotate through the daily challenges to keep the difficulty in the moderate range the trials used.