Young Adults Report the Most Loneliness and the Least Isolation. That Is the Whole Puzzle

Gallup’s 142-country survey, fielded from June 2022 to February 2023, found adults aged 19 to 29 reporting the highest rate of loneliness of any age group: 27%, against 17% among those 65 and older. That ordering shows up again and again. Harvard’s Making Caring Common survey put serious loneliness at 61% among Americans aged 18 to 25, against a national figure of 36%. Cigna’s 2025 report found Gen Z and Millennials lonelier than Gen X and Baby Boomers, despite having grown up with more communication tools than any prior cohort. A February 2026 eight-country study from researchers affiliated with Washington University in St. Louis reports that nearly half of young adults report loneliness, and that the association with depression and anxiety holds after adjusting for demographics and prior diagnoses.

None of this is in dispute. What is missing from almost all of this coverage is the other half of the picture: on the standard sociological measures of isolation — network size, frequency of contact, embeddedness in groups and institutions — young adults are not the most isolated age group. Older adults are. The National Academies’ 2020 consensus report put isolation at roughly one in four adults 65 and older, driven by retirement, bereavement, mobility limits, and the shrinking of social roles that comes with age. Young adults, by contrast, are more likely to live with roommates, attend school or work in group settings, and have larger nominal social networks than people in their seventies.

So the two curves cross. Isolation rises with age; loneliness, in recent data, does not. That crossing is the actual finding, and it gets flattened almost every time loneliness is reported as a single number.

Two different things, measured two different ways

Isolation is a property of a network: how many people are in it, how often contact happens, whether a person is embedded in groups. It can be counted from the outside — number of confidants, frequency of visits, membership in organizations. Loneliness is a subjective state: the gap a person feels between the connection they have and the connection they want. Cacioppo’s work on loneliness treats it as an aversive internal signal, akin to hunger, that can fire even when a person is objectively surrounded by others.

Holt-Lunstad’s 2015 meta-analysis in Perspectives on Psychological Science is one of the few sources to quantify both constructs against the same outcome and show they move separately. Social isolation carried an odds ratio of 1.29 for early mortality, loneliness 1.26, and living alone 1.32 — three related but distinct measures, each independently predictive, none reducible to the others. A 2024 study in Scientific Reports on isolation and loneliness during the pandemic makes the same point directly: the relationship between the two constructs varies by age, and that variation is the reason a single loneliness number obscures more than it reveals.

This matters for young adults specifically because the standard isolation metrics were largely built around retirement, widowhood, and mobility loss — the loneliness of the old. Applied to a 24-year-old with a full contact list and a shared apartment, those metrics register as fine. The self-report loneliness measures ask a different question — do you feel understood, do you feel you have people to rely on — and it is on that question that young adults score worst. The American Enterprise Institute’s 2021 survey on American friendship gives one mechanism: the share of Americans reporting no close friends rose from 3% in 1990 to 12% in 2021, and among men specifically from 3% to 15%. Network size held up reasonably well by some measures even as the number of people someone would call a close friend collapsed. Quantity of contact and quality of confiding relationship are not the same variable, and a survey that only asks about one will systematically misdescribe a generation defined by the other.

Why this gets conflated anyway

Part of the problem is instrumentation. Cigna’s index, the UCLA Loneliness Scale used in AARP’s surveys, and Gallup’s single-item “did you feel lonely a lot of yesterday” question are not measuring the same construct with the same precision, and none of them ask about network structure at all. A BMC Public Health review from 2023 names this directly: inconsistent measurement across the loneliness and isolation literature is one of the main barriers to comparing findings, and most large surveys default to asking about feeling rather than structure because feeling is cheaper to ask about in a single item.

The other part is narrative convenience. “Young people are the loneliest generation despite constant connectivity” is a clean sentence. “Young people report high loneliness on subjective scales despite low isolation on structural ones, which suggests the deficit is in the quality of contact rather than its quantity” is not a headline, but it is closer to what the data actually support, and it points toward a different set of interventions. If the problem were structural isolation, the fix would be about proximity and access — more third places, more shared physical settings. If the problem is a felt gap between actual and desired intimacy in a network that is nominally large, the fix is about the depth and reliability of specific relationships, which a bigger network does not automatically supply.

What would actually settle this

Most of the sources here are cross-sectional: a single snapshot compares age groups at one point in time, which cannot show whether an individual’s loneliness and isolation move together or apart as they age. A study that followed the same cohort of young adults over a decade, measuring both a structural isolation index and a subjective loneliness scale at each wave, would show whether the two converge with age — as the National Academies’ focus on older adults implies they eventually should — or whether a generation now reporting high loneliness with low isolation carries that mismatch forward as its network shrinks later in life. That is a materially different long-term public health question, and the current literature, largely single-wave and single-instrument, cannot answer it.