A large UK Biobank study of over 67,000 older adults found that lifetime cannabis users generally performed as well or better on cognitive tests than non-users, but early-onset and long-duration use were linked to worse outcomes in some domains. The findings are observational, lack pharmacological detail, and cannot establish that cannabis either helps or harms the aging brain.

Cannabis use among adults over 60 is increasing across jurisdictions where access has expanded, yet clinicians have remarkably little rigorous evidence to guide conversations about cognitive risk or benefit. Many older patients ask directly whether cannabis might protect their brain or whether past use has caused lasting harm. This study, drawing on one of the world’s largest health cohorts, represents the most ambitious observational attempt to answer that question to date, making its strengths and limitations critical for any clinician advising this growing patient population.

As cannabis legalization expands, clinicians face growing questions from older patients about the cognitive consequences of past, current, or prospective cannabis use. The underlying study by Sznitman and colleagues leveraged the UK Biobank, one of the world’s largest population health datasets, to examine associations between self-reported cannabis use history and cognitive performance across four domains: attention and executive function, processing speed, visual memory, and working memory. Of the cohort, approximately 17% reported lifetime cannabis use, and 4% of those had used in the past year. The study employed both cross-sectional analysis of the full 67,713-participant sample and longitudinal follow-up of 52,002 participants to capture cognitive trajectories over time.

In cross-sectional analysis, lifetime cannabis users performed better across all four cognitive domains compared to non-users. Current use was associated with better working memory, and past use was linked to slower decline in executive function in some longitudinal analyses. However, the findings were decidedly mixed: individuals who began using before age 17 showed poorer working memory as former users, and longer cumulative duration of use was associated with worse outcomes in certain domains. Critically, no effect sizes, confidence intervals, or p-values are reported in this secondary summary, and the study collected no data on dose, cannabinoid profile (THC versus CBD), route of administration, or medical versus recreational purpose. The authors themselves note that these exposure gaps, combined with recall bias and the observational design, prevent causal conclusions. The study team calls for prospective research with pharmacologically characterized exposures.

Cannabis and the Aging Brain: A Large Study Finds Mixed Results, Not a Clear Win

A study of over 67,000 older adults sounds like it should settle the question of whether cannabis harms the aging brain. It does not, and understanding why reveals something important about how we read observational research. The Sznitman team deserves genuine credit for assembling the largest dataset yet applied to this question and for publishing results that are honestly mixed rather than selectively positive. The UK Biobank’s scale and longitudinal component are real methodological assets. But the central problem is one of exposure definition: lumping all cannabis users together regardless of what they consumed, how much, how often, and through what route is roughly equivalent to studying “alcohol drinkers” as a single group, combining a person who drinks a glass of wine weekly with someone consuming a bottle of spirits daily, and concluding that “alcohol drinking” has a particular effect on liver health. The variable called “cannabis use” in this study is pharmacologically blind, and that blindness cascades through every association the study reports.

What makes the mixed findings more instructive than the headline finding is what they suggest about hidden heterogeneity. The fact that early-onset use and longer duration of use were linked to worse cognitive performance in specific domains tells us that the aggregate positive association is almost certainly masking real subgroup differences. This is a familiar pattern in large observational studies: when you average together groups whose underlying exposures differ markedly, you get a reassuring mean that conceals genuinely divergent trajectories. Add to this the problem of healthy survivor bias, because surveying cognitively intact older adults about their past drug use is a bit like surveying marathon runners at the finish line and concluding that running is universally safe, while ignoring everyone who dropped out, got injured, or never started. Those who experienced cognitive harm from cannabis may have been less likely to enroll, less likely to complete cognitive testing, or less likely to report past use at all. The trade journal framing of this study further compounds the interpretive challenge; by highlighting the positive associations in its headline and opening paragraphs, it nudges readers toward a conclusion the data simply do not support.

In my clinical practice, I would tell a patient that this study is modestly reassuring: it does not support the view that any history of cannabis use dooms the aging brain. But I would be clear that it is not evidence of benefit, and that the subgroup signals of harm from early and prolonged use are important to take seriously. To a colleague, I would say this is the best descriptive epidemiology we have on this question, but it cannot guide product-specific or dose-specific recommendations until we have data distinguishing THC-dominant from CBD-dominant products, edibles from inhalation, and daily use from occasional. To a policymaker, I would argue this study justifies continued research investment and, crucially, regulatory frameworks that require detailed pharmacological tracking rather than binary use-or-nonuse classification. The most durable lesson here is one that applies far beyond cannabis research: when an observational study’s headline finding is positive but its subgroup data are mixed, the subgroup heterogeneity, not the aggregate, usually carries the most scientifically important signal, because it reveals where the real questions lie.

For clinicians counseling older adults about cannabis, this study sits at an important but early position in the research arc. It represents a meaningful upgrade from the smaller, predominantly younger-adult studies that have characterized most of the cannabis-cognition literature. The dual cross-sectional and longitudinal design using a well-validated cohort provides more robust descriptive epidemiology than most prior work. However, the study falls well short of the pharmacological precision needed for clinical decision-making, and the trade journal summary through which most readers will encounter these findings omits the statistical detail necessary for independent appraisal.

From a pharmacological standpoint, the absence of data on cannabinoid ratios, dosing, and route of administration is the single greatest barrier to clinical translation. THC and CBD have distinct and sometimes opposing neurological effects, and grouping all forms of cannabis use into one variable obscures these differences entirely. Clinicians should also remain attentive to potential drug interactions in older adults who may be using cannabis alongside polypharmacy regimens, as these were not assessed. The one concrete recommendation supported by this evidence is that clinicians should not use this study to either discourage or encourage cannabis use for cognitive purposes, but should instead document patients’ cannabis histories with greater pharmacological specificity, including product type, cannabinoid content, frequency, and route, to contribute to the more granular data the field urgently needs.

This document is a trade journal news article, a secondary summary of a primary observational study published in a peer-reviewed journal (Age and Ageing). It carries no independent evidentiary weight and occupies the lowest tier of the evidence hierarchy. The underlying study itself is observational, using cross-sectional and longitudinal cohort data, which can identify associations but cannot establish causation. The single most important constraint on any inference drawn from this article is that it provides no statistical detail, meaning readers cannot independently assess the significance, magnitude, or robustness of any reported finding.

The broader literature on cannabis and cognition in older adults is sparse and methodologically limited, making this UK Biobank study one of the most substantial contributions to date. Prior work has largely relied on smaller convenience samples and younger populations. The finding that aggregate cannabis use is not associated with worse cognition is broadly consistent with some prior observational studies, though the mixed subgroup findings, particularly the association between early-onset use and poorer working memory, align with developmental neuroscience literature suggesting that adolescent cannabis exposure may carry unique risks. A small body of interventional work, including preliminary trials of THC-CBD extracts in dementia populations, explores a different question entirely: whether cannabinoids can treat existing cognitive decline rather than whether historical use predicts cognitive trajectories. These are complementary but fundamentally distinct research programs, and conflating them would be a significant interpretive error.

The most consequential analytic choice in the underlying study appears to be the treatment of “cannabis use” as a single undifferentiated exposure variable. Had the investigators been able to stratify by cannabinoid profile (THC-dominant versus CBD-dominant), route of administration, or dose intensity, the aggregate positive association could easily have fractured into distinct and potentially opposing patterns. Given the very large sample size, even trivially small cognitive differences would likely reach statistical significance, making the absence of effect size reporting in this summary particularly problematic. A sensitivity analysis excluding participants with unmeasured confounders, such as concurrent use of other substances or psychoactive medications, could also substantially alter the direction and magnitude of reported associations.

The most likely overinterpretation is that this study demonstrates cannabis improves or protects cognitive function in older adults. It does not. The observational design means that the better cognitive performance seen in lifetime cannabis users could readily be explained by confounding factors: cannabis users in the UK Biobank cohort may differ from non-users in education, socioeconomic status, physical activity, social engagement, or other health behaviors that independently predict better cognition. Healthy survivor bias further complicates matters, as individuals who experienced cognitive harm from cannabis may be systematically underrepresented in the studied cohort. Additionally, the trade journal publication venue should not be confused with peer-reviewed evidence; this news summary lacks the methodological detail needed for independent assessment and likely reflects framing favorable to cannabis.

This study contributes the largest observational dataset yet on cannabis use and cognitive aging, and its mixed findings are more informative than any single headline. It does not establish that cannabis protects, improves, or reliably harms cognition in older adults. For clinical practice, it offers modest reassurance that historical cannabis use is not uniformly associated with cognitive decline, while simultaneously flagging early-onset and long-duration use as patterns warranting further investigation. Clinicians should consult the original peer-reviewed paper, not this trade journal summary, before incorporating these findings into patient discussions.

Does this study prove that cannabis is good for the aging brain?

No. The study found associations between cannabis use history and cognitive test performance, but associations are not the same as proof of benefit. People who used cannabis may differ from non-users in many ways, including education, lifestyle, and overall health, that could independently explain better cognitive scores. The study design cannot separate the effect of cannabis from these other factors.

Should I start using cannabis to protect my memory as I age?

This study does not provide evidence to support starting cannabis use for cognitive protection. The researchers did not study whether initiating cannabis use improves cognition; they only examined associations between past use and current cognitive performance. Furthermore, some patterns of use, particularly starting young or using for a long time, were linked to worse outcomes in certain cognitive areas.

I used cannabis in my youth. Should I be worried about my brain health now?

This study is actually modestly reassuring on that point: overall, people with a history of cannabis use did not perform worse on cognitive tests than those who never used. However, those who started before age 17 showed some weaker performance in working memory, and longer cumulative use was linked to poorer results in certain areas. If you have concerns about your cognitive health, discuss your full history with your physician.

Why can’t the study tell us more about what type of cannabis matters?

The UK Biobank questionnaire asked participants whether they had used cannabis, when, and for how long, but did not collect information about what products they used, the THC or CBD content, how they consumed it, or at what dose. Because different cannabis products can have very different effects on the brain, this gap makes it impossible to know which specific exposures might be helpful, harmful, or neutral.

References

1. Sznitman SR, Vered S, Weinstein G. History of cannabis use and cognitive function in older adults: findings from the UK Biobank. Age and Ageing. 2025;54(11). DOI: 10.1093/ageing/afaf319

2. Colli M. Cognitive Improvements Witnessed with THC-CBD Cannabis Extracts and Dementia Patients. Cannabis Science and Technology. November 10, 2025. https://www.cannabissciencetech.com/view/cognitive-improvements-witnessed-with-thc-cbd-cannabis-extracts-and-dementia-patients

3. McEvoy E. Updates on Cannabis Research for Alzheimer’s Disease Symptoms. Cannabis Science and Technology. November 20, 2025. https://www.cannabissciencetech.com/view/updates-on-cannabis-research-for-alzheimer-s-disease-symptoms

Have thoughts on this? Share it: