Do friendships help you live longer? What the research actually shows

The figure everyone quotes comes from a 2010 meta-analysis: 148 studies, 308,849 people, an average of seven and a half years of follow-up, and a 50% higher likelihood of survival for those with stronger social relationships. It held across age, sex, starting health and cause of death. The authors compared the size of the effect with the well-known risk factors and found it comparable to quitting smoking and larger than obesity or inactivity.

Two things about that number are usually left out. First, it is not about loneliness. The strongest results came from rich measures of how integrated someone was, how many roles and ties and kinds of support they had, and the weakest from crude ones such as whether they lived alone. Second, the “fifteen cigarettes a day” line that grew out of it is a comparison of effect sizes across different studies, not an experiment, and the researcher behind it keeps a page correcting how it gets used.

A 2023 meta-analysis of 90 cohort studies and 2.2 million people updated the picture and split it in two: social isolation went with a 32% higher risk of dying from any cause, loneliness with 14%. Both matter. But having people around you carries roughly twice the association that feeling alone does, which is worth knowing if you are deciding what to do about it.

Most of the older work lumps every tie together. Two recent studies pull friendship out on its own. In the Health and Retirement Study, 12,998 US adults over 50 were scored on friendship strength, a mix of how many close friends they had, how often they were in touch and whether those friendships were a support or a strain. The top quarter had a 24% lower risk of dying over the next eight years than the bottom quarter, along with less depression, less loneliness and a lower risk of stroke.

A 2025 analysis of the MIDUS cohort asked which relationship’s quality predicts lifespan. Friendship came first. Family ties mattered to a lesser degree, and a strained relationship was more closely tied to early death than an unsupportive one. That last part matters for anyone tempted to read this as “see more people”: the finding is about quality, and a relationship that drains you is not on the good side of the ledger.

Set that next to what we know about how relationships decay, and something uncomfortable falls out. Friendship is the tie most associated with a long life, and friendship is also the one that fades when contact stops, while family closeness survives the same silences. The relationship with the most riding on it is the one nobody maintains for you.

The largest single study on this followed 458,000 people in UK Biobank for twelve years. Of five measures of social connection, how often friends or family visited had the biggest association with mortality. People who were never visited had a 39% higher risk of dying over the follow-up than those visited daily. Visits every three months carried 11%.

The useful part is the shape. Below monthly, risk climbed. At monthly or more, it flattened, which the authors describe as a threshold. This is not a dose curve where every extra coffee buys a little more; it is a floor, and the floor is lower than most people’s guilt sets it. Not weekly. Once a month.

The combinations were the sharpest finding. Living alone on its own carried a 25% higher risk. Living alone and never being visited carried 77%. Living alone is survivable. Living alone unvisited is what the association is really about.

That any of this is causal. Every study here is observational, and the plausible reverse story is strong: illness reduces social contact long before it reduces lifespan, so some of what looks like friendship protecting health is health enabling friendship. No trial has changed people’s friendships and measured what happened to their mortality, because you cannot randomise people into having friends.

A 2024 study of 476,100 people tested the loneliness half of this genetically. Observationally, loneliness was linked to 30 of 56 diseases. Under Mendelian randomization, most of those links did not hold, and the authors concluded loneliness is more often a marker of illness already present than a cause of it. That does not touch the isolation findings, which were stronger to begin with, but it is why this article says “linked to” and never “causes”.

That friends are medicine. In the same study that found the 24% lower mortality, the strongest friendships also went with more smoking. Friends carry habits both ways. And that a number would help: nothing here supports scoring your relationships, and the research on measuring enjoyable activities says it makes people enjoy them less. A monthly quota is a score.

Treat the monthly line as permission, not a target. The floor in the data is low. If there is one friendship you have let slide past it, the research says that is the one worth a message this month, and the same research says the message will land better than you expect: people who reach out after a silence consistently underestimate how pleased the other person is.

Spend the effort on quality, not count. Two of the three friendship studies here score quality as support versus strain, and the third finds quality outranks headcount entirely. The cheapest way to make a friendship feel supportive from the other side is to remember what they told you and ask about it next time. Feeling understood is the working definition of closeness, and a remembered detail is how you deliver it.

None of this needs an app. A note on your phone with a name and one line works. What does not work is intending to remember, because that is precisely the thing that fails, and the friendship that lapses is the one you meant to get back to.