← Back

The longest study of happiness

Social Fitness: The Longevity Habit You Skip

Harvard has followed the same lives since 1938. The finding that keeps beating cholesterol, IQ, and income is not a personality type. It is whether you keep your relationships in training.

longtime.live 23 Aug 2026 11 min read Sources

1938Year the Harvard Study of Adult Development began
724Original men: Harvard sophomores plus young men from inner-city Boston
1,300+Descendants now followed as the study enters its third generation
85+ yrsThe longest in-depth study of adult happiness and health

People treat friendships like furniture. Once they are in the room, they assume they will stay there. The longest study of adult lives ever run says that is the mistake.

Harvard’s Study of Adult Development has been asking a blunt question since 1938: what actually predicts a good life? Not the life people advertise. The one they are still living in their seventies and eighties, with their health, their mood, and someone who would notice if they disappeared.

The answer that keeps surviving the next decade of data is not fame, income, or a better cholesterol number at fifty. It is the quality of close relationships, practised on purpose. Robert Waldinger and Marc Schulz, the study’s current directors, gave that practice a name that sounds almost too ordinary: social fitness.

This is the companion to our piece on happiness and longevity. That article asks whether feeling good extends life. This one asks what the longest happiness study says you should do with other people, what you should stop doing, and why the mortality literature treats isolation like a physical risk factor.

What the longest study actually is

In 1938, Harvard researchers enrolled 268 college sophomores in what became the Grant Study. A second cohort, 456 young men from some of Boston’s poorest neighbourhoods, joined through the Glueck Study. Together they made 724 original participants. Researchers collected medical exams, blood, brain scans, and pages of interviews about work, marriage, friendship, and worry, then kept going as the men aged, married, raised children, retired, and died.

The study now includes more than 1,300 descendants. It is not a random sample of humanity. It began with men, many of them white, in one American city. That limit matters, and it is why the Harvard findings are strongest when other groups later find the same pattern.

They did. Satisfaction with relationships at midlife predicted who was happy and healthy at eighty better than midlife cholesterol did. George Vaillant, who directed the study for decades, compressed the result into a line that still holds: the key to healthy ageing is “relationships, relationships, relationships.”

The Harvard Study of Adult Development at a glance A timeline infographic of the Harvard Study of Adult Development. It starts in 1938 with 724 men, follows them through midlife health measures, finds that relationship satisfaction at age 50 predicts health at 80 better than cholesterol, and now includes more than 1,300 descendants after 85-plus years. Harvard Study of Adult Development The longest in-depth study of adult happiness and health 1938 Study starts 724 young men Midlife Health recorded cholesterol, blood pressure, marriage Age 80 Look back relationships at 50 beat cholesterol Now Still running 1,300+ descendants 85+ years of data What they measured, again and again Medical exams, blood, later brain scans, and detailed interviews about work, marriage, friendship, and worry. The surprise was not that close ties made people happier. It was that those ties also predicted who stayed healthier. Waldinger and Schulz call the maintenance of those ties social fitness: a practice, not a personality trait.
Study snapshot from the Harvard Study of Adult Development and Waldinger and Schulz's reporting in The Good Life. The original cohorts were men; later waves added partners and descendants.

The key findings

The Harvard team did not set out to prove that friendship is medicine. They expected the usual midlife numbers, lipids, blood pressure, smoking, to do most of the predictive work. When they looked back from age eighty, relationship satisfaction was the stronger signal.

That finding sounded soft, so they waited for other labs. The same pattern showed up: people with warmer, more reliable connections were less likely to develop coronary disease, type 2 diabetes, and arthritis, and they recovered faster when they did get sick. Late-life secure attachment to a partner has been linked to better brain health a few years on. Childhood warmth even left a faint mark on how securely people attached to a partner six decades later.

A few results are easy to misuse. Here is the precise version.

Quality beats quantity

A crowded contacts list is not the finding. Feeling you can count on someone is. Several people help. One solid person is the minimum.

One night-time call

Researchers asked who you could ring if you were sick or scared at 3am. Most people listed several names. Some, including some who were married, listed nobody.

Fitness, not furniture

Good friendships wither from neglect. You do not finish a gym session and declare yourself fit for life. Social life is the same kind of living system.

Status is a weak bet

Once basic security is met, extra wealth, titles, and badges add little happiness. Meaningful work helps. The plaque on the wall does not.

It is not too late

People in the study found a tribe in retirement, or love in their seventies, after decades of thin connection. The data do not support writing yourself off.

Weak ties still count

Family can be enough for some people. Others get bits of well-being from colleagues, neighbours, and the person who makes the coffee. Repeated casual contact is how deeper ties start.

The practical coinage is social fitness. Waldinger’s analogy is the gym. Nobody finishes Tuesday’s workout and decides they never have to move again. Plenty of people treat a good marriage, or a college friendship, as a completed project. The study’s happiest older adults were the ones who kept reaching out, making plans, and repairing the ordinary ruptures that every long relationship collects.

How this ties to longevity

If this were only a mood finding, it would still be worth the time. It is not only a mood finding.

A 2010 meta-analysis by Holt-Lunstad, Smith, and Layton pooled 148 prospective studies and 308,849 people. Stronger social relationships were associated with a 50% higher likelihood of survival (odds ratio 1.50). Studies that used richer, multidimensional measures of social integration found an even larger association: a 91% higher likelihood of survival (odds ratio 1.91). The effect held across age, sex, baseline health, and cause of death. The authors judged it comparable to quitting smoking and larger than the usual estimates for obesity and inactivity.

A later 2015 meta-analysis, covering more than 3.4 million people, looked at the deficit side. After adjustment, loneliness was linked to a 26% higher likelihood of death, social isolation to 29%, and living alone to 32%. Feeling lonely and being isolated were both risky. They are not the same thing, and solitude is not automatically a problem. The harm shows up when aloneness is unwanted, chronic, and leaves you without a recovery system.

That is why the US Surgeon General’s 2023 advisory treated social connection as a public-health issue, not a lifestyle extra. The often-quoted comparison with smoking 15 cigarettes a day is a visual approximation of premature-mortality odds, not a claim that loneliness delivers nicotine. We have already taken that metaphor apart in the Cigarette Index. Use it as a severity cue. Do not use it as a calculator.

Social connection and survival A bar chart of odds ratios for survival from Holt-Lunstad 2010: 1.91 for complex social integration, 1.50 for stronger social relationships overall, 1.46 for functional support, and 1.19 for living with others versus living alone. ODDS OF SURVIVAL VERSUS WEAKER CONNECTION 1.00 1.45 1.91 Complex social integration 1.91 All social-relationship measures 1.50 Functional support 1.46 Not living alone 1.19 Holt-Lunstad, Smith and Layton, PLOS Medicine 2010. Odds ratios above 1 mean higher odds of being alive at follow-up.
Richer measures of social integration predicted survival more strongly than a simple "lives alone" checkbox. These are observational associations, not assigned treatments.

The Harvard team’s working mechanism is stress recovery. A hard day puts the body into fight-or-flight: heart rate up, blood pressure up, stress hormones in circulation. If you can take that load to a person who actually listens, the body can stand down. If you cannot, the system stays slightly on. Over years, that looks like more inflammation, worse sleep, higher cardiovascular risk, and less motivation to do the boring protective things.

Social connection also changes behaviour. People with someone expecting them show up for walks, take the medication, go to the appointment, and eat a meal that is not just whatever is nearest the couch. In the Million Veteran Program, social connections sat alongside sleep, movement, and not smoking as one of eight lifestyle factors tied to large gaps in modelled life expectancy. Two of those eight were about the mind and other people.

How a good conversation gets into the body An infographic of the stress-recovery loop. A daily stressor triggers fight-or-flight. Talking with a trusted person helps the body return to equilibrium. Without that person, low-level stress hormones and inflammation persist and wear on health over time. Stressor ordinary hard day Fight-or-flight heart, hormones up A real listener body stands down Recovery back to baseline If there is nobody to call The stress response stays slightly on. Circulating stress hormones and inflammation do not fully clear. Over years that wear shows up in sleep, blood vessels, immunity, and the energy to keep the rest of a longevity routine going. Isolation is the missing recovery step, not a separate mystical risk.
Waldinger's working model: relationships regulate stress. The physiology is still being mapped, but the direction of the evidence is consistent across labs.

What to do

Social fitness is not a charisma transplant. It is a set of small, repeatable actions. Waldinger’s first move is almost insultingly small: think of one person you miss, then send the text.

The study’s other practical tool is an inventory. Map the people in your life against seven kinds of support. You do not need one person to do all seven. You do need the set to be covered by someone, and you need to notice the cells that have gone empty.

Seven keystones of support A seven-cell infographic of Waldinger and Schulz's keystones of support: safety and security, learning and growth, emotional closeness, identity and shared experience, romantic intimacy, practical help, and fun and relaxation. Seven keystones of support From Waldinger and Schulz, The Good Life. One network can cover these. One person usually cannot. 1. Safety Who do you call at 3am if you are scared or sick? Crisis coverage, not vibes. 2. Growth Who pushes you to try, learn, or take a chance? Ambition needs witnesses. 3. Confiding Who knows the real story when you are low? Honesty, not performance. 4. Identity Who has been there for the long shared plot? Someone who knows you. 5. Intimacy Are you satisfied with closeness, not just status? Optional in form, not need. 6. Help Who answers a practical problem, not just a feeling? Wi-Fi, lifts, expertise. 7. Fun and ease Who makes you laugh, rest, or want the next walk? A life with only crisis contacts is still a thin life. Play is a health behaviour here, not a leftover.
Use the keystones as a gap check, not a score. Notice who energises you, who depletes you, and which cell has nobody in it.

Then do the unglamorous maintenance.

  • Send the small signal. A text, a voice note, a two-line email: I was thinking of you. Waldinger’s observation is that the reply rate is higher than proud people expect.
  • Put the contact on a calendar. Walks, coffee, a recurring call. Relationships that depend on “we should soon” die of scheduling.
  • Join something you already care about. A club, a choir, a volunteer shift, a campaign, a pickup game. Shared work is the least awkward way to meet the same people twice.
  • Keep the weak ties warm. The neighbour, the barista, the colleague you like. Those micro-conversations are not fake social life. They are part of the dose.
  • Be the listener you would want. Attention is the scarce resource. Phones on the table tax it. Curiosity is a trainable social muscle.
  • Repair after the ordinary rupture. Every close tie collects slights. The WISER sequence in The Good Life is watch, interpret, select, engage, reflect. The skill is not never fighting. It is not letting the fight become the relationship.
  • Offer the seven keystones back. Support that only flows in one direction is not a network. It is a service.

Introverts do not get a pass, and extroverts do not get extra credit. Shy people often need fewer relationships; they still need one they can count on. The prescription is coverage, not a busier Saturday.

What not to do

The study is just as clear about the traps. Most of them look like competence from the inside.

Social fitness: do and do not A two-column infographic. The do column lists reaching out, repeating contact, covering support keystones, and staying present. The do-not column lists neglecting ties, replacing people with feeds, waiting for retirement, expecting one person to do everything, and isolating under stress. Do Reach out in tiny, specific ways See the same people more than once Cover the keystones across a network Keep some friction in leisure Put the phone away for the talk Repair after ordinary conflict Use groups built around a shared task Treat casual contact as real contact Ask who you would call at 3am Start again after a thin decade Maintenance looks small in the moment. Compounded, it is how people stay known in their eighties. Do not Assume good friends maintain themselves Swap people for a feed or a streak Wait until retirement to rebuild Expect one partner to be the whole net Isolate because you are busy or low Chase status as the happiness plan Confuse solitude with loneliness Keep only depleting, high-status ties Treat social life as a finished project Decide it is too late for you Most of these feel efficient. They are how a social life quietly thins while the calendar stays full.
The don'ts are mostly neglect with a good excuse. Career intensity, screens, and "I will sort my life later" are the usual ones.

A few of those traps deserve a longer warning.

Do not outsource the whole job to a spouse. Marriage can be a powerful health relationship. It is not automatic cover. Some of the study’s married participants still could not name a night-time call. One person cannot be safety, play, growth, and practical help every week without burning out.

Do not wait for the free time. The retirement literature on this site already argues against treating later life as a sudden social project. Workplaces and schools manufacture contact. When they stop, the people who already had a practice keep going. The people who postponed it discover that a calendar full of errands is not a community.

Do not confuse a feed with a friend. Phones are useful for the text that starts a walk. They are a weak substitute for the walk. Waldinger and Schulz are blunt about attention: if you are half in the conversation and half in the glass, you are practising a thinner form of the thing that predicts health.

Do not isolate because you feel low. That is the moment the recovery system is for. Withdrawal is understandable and still costly.

What this cannot promise

Read the limits
  • It is observational. Nobody was randomised to a better marriage. People who keep friends also differ in a hundred other ways.
  • The original sample is narrow. The first cohorts were men in one city. Later waves and other labs broaden the picture. They do not make the Harvard men a stand-in for everyone.
  • Reverse causation is live. Illness shrinks social life. Some of the mortality gap is sickness causing isolation, not only isolation causing sickness. The community-cohort analyses make that less likely as the whole story. They do not erase it.
  • Years are not for sale. A 50% higher survival odds is not a coupon for a decade. It is a population association. Our [editorial policy](/editorial-policy/) is to keep that distinction visible.

None of that makes the result ornamental. A signal this stubborn, across 85 years of one study and hundreds of others, is exactly what you would expect if close ties were doing biological work. It is not a reason to abandon sleep, movement, or blood pressure. It is a reason to stop filing other people under “personal life” while you optimise everything else.

The bottom line

Social fitness is the unfashionable longevity habit: keep your relationships in training. The longest study of happiness says the people who age well are not the ones with the best midlife labs. They are the ones who stay known.

Send the text. Walk with someone. Cover the keystones across a few people, not one overloaded partner. Do not wait for retirement, a less busy quarter, or a more sociable version of yourself. The body keeps the score of whether stress has somewhere to go.

Want a rough map of how the rest of your habits stack up? Our longevity calculator turns diet, sleep, movement, and the usual risks into a life-expectancy estimate.

Sources