Friends in Later Life: Networks Narrow at the Edges as Close Ties Hold
Published 2026-11-15
Social networks tend to get smaller in later life, but the research finds that most of what falls away is the outer circle of acquaintances; the closest ties tend to hold[1][2]. In a 2023 US survey, adults aged 65 and over were more likely than any younger age group to say they had five or more close friends[3].
The picture of old age as a slow slide into isolation does not match what the large surveys show. Older people lose colleagues, friends and partners, and they also keep old friendships going, make new ones, run clubs, worship and volunteer. This study sets out what surveys and cohorts in the United States, Europe and China show about how networks change with age, how friends and family differ, and what joining groups is linked with. Network size and contact are measures of social isolation, not of loneliness: a small network is not the same as feeling lonely, and we keep the two apart throughout.
Networks shrink with age, but mostly at the edges
The widest summary of the evidence is a 2013 meta-analysis of 277 studies with 177,635 participants, from adolescence to old age. Both cross-sectional and longitudinal studies found that people’s overall social networks grew until young adulthood and then “decreased steadily”, that personal and friendship networks shrank throughout adulthood, and that family networks stayed about the same size from adolescence to old age. Life events tied to particular ages, such as widowhood, brought network changes like those seen with age itself[1]. The authors also found a period effect: personal and friendship networks had become smaller over the previous 35 years[1].
Which ties are lost matters. A small US study, which began with 184 adults aged 18 to 94 in the San Francisco Bay area and returned to them five and ten years later, found that, as networks shrank, “reductions were observed primarily in the number of peripheral partners; the number of close partners was relatively stable over time”. Older participants also said the people in their networks brought them more positive and less negative emotion[2]. The sample was small and relatively healthy, and 60 people completed all three rounds. The researchers read the pattern through socioemotional selectivity theory, the idea that people invest more in close ties as they age. That is a theory, not something the study tested directly: illness, bereavement and lost workplaces or transport can also thin a network without anyone choosing it.
The US National Social Life, Health, and Aging Project (NSHAP) gives a national profile. Its 2005–2006 survey of 3,005 Americans aged 57 to 85 asked people to name those they had discussed important matters with, up to five, and to say if there were more. The average network held 3.6 people, and the most common answer was 5[5]. Older age was linked with smaller networks, less closeness to network members and fewer ties outside the immediate family; each ten years of age went with a 6 percent increase in the number of network members who were a spouse or children[5]. But the oldest respondents were not less close to their networks when they saw them often: the link between age and lower closeness was strongest among people with infrequent contact and “virtually non-existent” among those with frequent contact[5]. The authors concluded that their findings were “inconsistent with the notion that old age has a universal negative influence on social connectedness”[5].
NSHAP surveyed people living at home, not in care homes, and a single survey compares generations rather than following anyone over time. Our study of loneliness across the life course looks at how loneliness itself, rather than network size, differs by age.
Many older people report a wide circle of close friends, though who counts as a confidant differs by country
In a Pew Research Center survey of 5,057 US adults in July 2023, people were asked how many close friends they had, not counting family. Of those aged 65 and over, 49% said five or more, compared with 40% of those aged 50 to 64, 34% of those aged 30 to 49 and 32% of those under 30[3]. Overall, 8% of US adults said they had no close friends; the published summary does not break that figure down by age. Among people with at least one close friend, those aged 50 and over were more likely than younger adults to be highly satisfied with their friendships (77% against 67%)[3].
This does not contradict the smaller confidant networks NSHAP found among its oldest respondents: confidants are a narrow measure and mostly family, close friends a looser one that leaves family out. Neither measures loneliness.
The Survey of Health, Ageing and Retirement in Europe (SHARE) asks people aged 50 and over to name the people they have “discussed important things” with, up to seven. An analysis of 56,093 respondents in 17 countries in 2016 found an average network of 2.6 people, and about 80% named three or fewer[4]. Family dominated these lists. Whether a friend appeared at all varied widely in the study’s unweighted sample: from 43.2% of respondents in Denmark and over 40% in Switzerland, Belgium and Sweden to 16.4% in Italy and 10.3% in Poland[4]. Neighbours appeared far less often than friends in every country. In the study’s adjusted models, neighbour ties were relatively more pronounced for older people without a partner or children in southern Europe and lower-GDP countries[4].
Friends matter most where family is absent. After adjusting for age, health, income and other characteristics, older people with neither a partner nor children were 23 percentage points more likely than those with both to name a friend as a confidant, and those without a partner but with children 14 points more likely[4].
Life changes reshape networks, and not always by shrinking them
Widowhood is the clearest example. A SHARE study of 55,555 Europeans aged 50 and over, 3,257 of whom were widowed between survey waves, found that losing a spouse was followed by sustained increases in network size, with more children and friends named, especially among men. Yet satisfaction with the network fell and stayed lower, and widowed women more than men took up voluntary work and sport or social clubs[6]. Our study of widowhood and loneliness covers what follows a partner’s death in detail.
Retirement removes the workplace, often the largest source of everyday acquaintances; our study of retirement and loneliness sets out what cohort studies find about that transition. NSHAP’s authors noted that retirement and bereavement may even “prompt greater connectedness” in community life[5].
Networks also turn over. Tracking confidants across 13 European countries over four SHARE waves, the last in 2021–2022, researchers found that the number of confidants was not associated with the pandemic, but older adults lost close network members much more during it than before, while also making new ties[7]. A stable count can hide real losses and real new friendships.
Friends and family are linked with wellbeing in different ways
Several large studies find that friendships carry particular weight for wellbeing in later life, alongside family rather than instead of it. None of the studies below measured loneliness; they measured life satisfaction, mood, health or wellbeing, and all are observational, so they show associations rather than cause and effect.
A 2000 meta-analysis of 286 studies of older people found that the quality of social contacts was more strongly associated with subjective wellbeing than the quantity. Contact with friends was more strongly related to wellbeing than contact with adult children, but the quality of contact with adult children was more strongly linked with life satisfaction than the quality of friendships[8].
A German national survey that followed 2,830 middle-aged and 2,032 older adults over time found that, in the older group, activities with friends were followed by more positive feelings and higher life satisfaction and by fewer negative feelings. Activities with family were followed by both more positive and more negative feelings[9]. The authors point to friendships being chosen and family ties obligatory.
A 2017 paper reported two studies. In a cross-sectional survey of 271,053 adults, valuing friendships was linked with better functioning on health and wellbeing measures, particularly among older adults, while the link with valuing family was much the same at every age. In a longitudinal study of 7,481 older adults, strain in friendships, and not in family ties, predicted more chronic illness over six years, and support from spouses, children and friends all predicted higher wellbeing over eight years[10]. Healthier people may simply find it easier to keep friendships going.
Worship and volunteering were more common among the oldest people surveyed, but what groups add is hard to separate from who joins
Community life is where the oldest people in the NSHAP survey stood out. These are differences between people of different ages in one survey, not change as people grow older. Older age was linked with more socialising with neighbours, more religious participation and more volunteering. After adjustment for health, marital history, retirement and other characteristics, the predicted chance of attending religious services at least weekly was 42% for the youngest respondents and 65% for the oldest, and the predicted chance of volunteering at least weekly was 20% at age 57 against 29% at age 85[5]. Meetings of organised groups were the exception: the oldest respondents were “no more likely” than the youngest to attend them often[5].
Whether taking part in groups is linked with less loneliness is a harder question, and the answers differ by activity. A US study using the National Health and Aging Trends Study surveyed 7,547 people with an average age of 75 in 2023 and 6,543 of them again in 2024. Continuing to volunteer over the two years was associated with lower loneliness in 2024, after taking account of loneliness in 2023, and so was taking part in a larger number of activities. Attending club meetings, attending religious services and going out for enjoyment were not significantly associated with loneliness[11]. The abstract does not describe how loneliness was measured. Our study of volunteering in later life looks at that evidence in full.
In China, a study of 6,758 middle-aged and older people followed over five waves of the China Health and Retirement Longitudinal Study (CHARLS), from 2011 to 2020, found that loneliness and social participation were linked in both directions over time. People with higher participation at the start tended to see loneliness fall more, or rise less, over time, while starting loneliness was not associated with later changes in participation[12]. Again, the abstract does not name the loneliness measure.
The common problem is selection. People who join groups often differ from those who do not in health, income, education and mobility before they join, and poor health or loneliness can make it harder to keep going. Adjusting for some of this cannot rule it out, so these studies show what membership goes with, not proof that joining a group reduces loneliness.
What the evidence asks of us
The WHO Commission on Social Connection, reporting in June 2025, separates the two things this study has kept apart: social isolation, “the objective lack of sufficient social connections”, and loneliness, “the painful feeling that arises from a gap between desired and actual social connections”[13]. It calls on governments and communities to make social connection “a public health priority”, and names “strengthening social infrastructure (e.g., parks, libraries, cafés)” among the solutions[13]. The friendship evidence points the same way:
- Keep the places where friendships happen open and reachable. Clubs, faith groups, u3a groups, lunch clubs and volunteering are where many older people keep and renew their ties, and the oldest people in the NSHAP survey were more likely than the youngest to worship and volunteer weekly, and no less likely to attend organised groups. Their meeting places and transport are social infrastructure, not extras.
- Plan for people without close family. In SHARE, older people without a partner or children leaned much more on friends. Services built around a spouse or adult child as the main contact can miss them.
- Pay attention after big transitions. Widowhood and retirement change who is in a network, and network satisfaction fell after widowhood even as networks grew. These are the points where the evidence on changing networks is clearest, and where contact from groups and services can matter most.
- Measure networks and loneliness separately. Counts of confidants, close friends and group memberships describe social isolation; they do not tell a service who feels lonely. Surveys and evaluations should collect both, with a named loneliness measure.
- Fund better evidence on groups. Most of it is observational and shaped by who joins; longer follow-up, comparison groups and the views of people who stopped attending would help.
Language: this piece says "older people" and treats loneliness as a common human experience that anyone can have, not a personal failing or an illness. It keeps apart three things that are often mixed up: loneliness (how people feel about their relationships), social isolation (how few social contacts they have) and living alone (a household arrangement). Where a study uses another term, such as "seniors" or "older adults", we keep it when quoting that study.
Why we're publishing this
Groups run by and for older people, from u3a groups and lunch clubs to older people’s charities, are often told that old age means shrinking, lonely lives. The research is more varied and more hopeful than that: networks narrow at the edges, close ties tend to hold, and in the US survey the oldest respondents were at least as active in community life as the youngest. We publish this so that those groups can cite the evidence accurately, including where it is thin, when they make the case for the places and activities that keep friendships going. It is part of our Loneliness in Later Life research library.
How to read this data
Most figures here measure social networks: the number of people someone discusses important matters with (NSHAP, up to five named; SHARE, up to seven), the number of close friends someone reports (Pew), or how often people take part in groups. These are measures of social isolation or connection, not of loneliness. The studies of friends and family measured wellbeing, life satisfaction, mood or health, also not loneliness. Only the US and Chinese studies of group participation measured loneliness, and their published abstracts do not say how.
Almost all the evidence is observational. Cross-sectional surveys (Pew, SHARE in 2016, NSHAP in 2005–2006) compare age groups at one time, so differences mix ageing with generational change. Longitudinal studies follow the same people but cannot rule out that health or income shape both friendships and wellbeing. Predicted percentages from models are not raw survey shares. The evidence comes mainly from the United States and Europe, with one Chinese cohort; our search did not turn up comparable network data from Africa, Latin America or most of Asia.
Use this data
Free to cite with attribution to Understanding Aging. u3a groups, men's sheds, senior centres, lunch clubs, befriending services, charities, campaigners and journalists are welcome to reuse these figures and charts. For the underlying figures as a spreadsheet, get in touch.
Important: This article summarises published research and statistics for general information. It is not medical or care advice and cannot be used to diagnose, treat or predict anything about an individual. If loneliness, low mood or isolation is affecting your health or someone else’s, a doctor or another health professional can help, and local services for older people can often put you in touch with groups and support nearby.
Sources
- Wrzus C, Hänel M, Wagner J, Neyer FJ. Social network changes and life events across the life span: a meta-analysis. Psychological Bulletin, 2013. europepmc.org/article/MED/22642230
- English T, Carstensen LL. Selective narrowing of social networks across adulthood is associated with improved emotional experience in daily life. International Journal of Behavioral Development, 2014. europepmc.org/article/MED/24910483
- Pew Research Center. What does friendship look like in America? Pew Research Center, 2023. pewresearch.org/short-reads/2023/10/12/…
- Mair CA. Friend and neighbor ties among unpartnered, childless, and “kinless” older adults in Europe. Population, Space and Place, 2026. pmc.ncbi.nlm.nih.gov/articles/PMC13596058
- Cornwell B, Laumann EO, Schumm LP. The social connectedness of older adults: a national profile. American Sociological Review, 2008. pmc.ncbi.nlm.nih.gov/articles/PMC2583428
- Tambellini E, Danielsbacka M, Tanskanen AO, Rotkirch A. Life after loss: gendered changes in social networks and community participation around widowhood. Journals of Gerontology, Series B, 2026. europepmc.org/article/MED/42490216
- Menashe-Oren A, Uccheddu D, Rizzi EL. Social ties in old age: the effect of the COVID-19 pandemic. European Journal of Ageing, 2025. europepmc.org/article/MED/41186783
- Pinquart M, Sörensen S. Influences of socioeconomic status, social network, and competence on subjective well-being in later life: a meta-analysis. Psychology and Aging, 2000. europepmc.org/article/MED/10879576
- Huxhold O, Miche M, Schüz B. Benefits of having friends in older ages: differential effects of informal social activities on well-being in middle-aged and older adults. Journals of Gerontology, Series B, 2014. europepmc.org/article/MED/23682165
- Chopik WJ. Associations among relational values, support, health, and well-being across the adult lifespan. Personal Relationships, 2017. doi.org/10.1111/pere.12187
- Choi NG, Marti CN. Types of social participation and self-reported loneliness among older adults: cross-sectional and longitudinal associations. Clinical Gerontologist, 2026. europepmc.org/article/MED/41934642
- Li Q, Chen Z. The dynamic association between loneliness and social participation among Chinese middle-aged and older adults. BMC Psychology, 2026. europepmc.org/article/MED/42449440
- World Health Organization. Social connection linked to improved health and reduced risk of early death (news release on the WHO Commission on Social Connection report). WHO, 2025. who.int/news/item/30-06-2025-…