You send your partner a headline: “One in six people feel lonely.” They look up from the sofa and say, “But you have me.” The number has barely arrived before it becomes a question about the two of you. Loneliness statistics can open a useful conversation. First, keep the people, years and meaning attached.
WHO’s 2025 report estimated average global loneliness prevalence at 15.8% across 2014 to 2023, approximately one in six. Its model combined people aged 13 and over, with a 95% uncertainty interval of 12.8% to 20.1%. This describes a population over a period, rather than counting who feels lonely tonight.
What the one-in-six figure actually counts
The WHO Commission on Social Connection report drew on 23 datasets covering 153 countries and territories. Annex 1 explains that statistical modelling combined the available evidence to estimate average prevalence across ten years. The combined age range starts at 13; the headline does not supply an estimate for younger children.
Keep the survey dates, estimate period and publication date apart. A new article repeating the finding does not make it newly measured. The decade is an averaging period, not ten years of following the same people or counting everyone who ever felt lonely.
Loneliness measures a gap, not the size of your household
WHO distinguishes loneliness, a distressing mismatch between wanted and experienced connection, from social isolation, which concerns too few relationships, roles or interactions. Being alone is different again. Someone can welcome a quiet evening and feel connected; someone with plenty of company can still miss the support they need.
A family chat pinging all afternoon shows contact. It cannot tell you whether anyone is available for a difficult conversation. A busy calendar or shared address cannot settle how connected someone feels. Ask about the experience instead of counting people nearby.
How different surveys become one estimate
Some datasets used several questions in a loneliness scale; others used a single question, with different recall periods or none specified. Annex 1 describes the definitions and adjustments bringing these sources together. The final figure is model-based, rather than the result of one worldwide poll.
Read 15.8% beside its 95% uncertainty interval, 12.8% to 20.1%. That interval is part of the reported result. Dropping it makes the decimal point look more exact than the evidence warrants. It is not a range for one person’s feelings or a threshold for deciding whether someone needs support.
WHO also describes uneven data coverage and too few repeated measurements to establish comprehensive global time trends. The period includes the pandemic; this average alone cannot show how much loneliness changed because of it. The limitations also warn that surveys generally could not distinguish transient from chronic loneliness. Do not turn the estimate into a count of people with a diagnosed condition.
Read age and income comparisons with their limits
Figure 2 gives these global age-group estimates for 2014 to 2023, with 95% uncertainty intervals:
- Ages 13 to 17: 20.9%, with an interval of 15.6% to 28.5%.
- Ages 18 to 29: 17.4%, with an interval of 13.2% to 23.6%.
- Ages 30 to 59: 15.1%, with an interval of 11.0% to 21.2%.
- Ages 60 and over: 11.8%, with an interval of 8.4% to 17.6%.
These broad bands do not predict an individual’s experience. WHO notes measurement differences and that grouping everyone aged 60 and over can obscure differences within later life. The estimates do not prove why the groups differ.
The report gives 24.3% for low-income countries, with a 95% uncertainty interval of 20.0% to 29.4%, and 10.6% for high-income countries, with an interval of 8.3% to 14.0%. These are country income groups, not an assessment of a particular household’s income or relationship.
A version you can repeat without changing the finding
For a post, presentation or conversation, keep the source and its scope in the same sentence:
“WHO’s 2025 report estimated average global loneliness prevalence among people aged 13 and over at 15.8%, approximately one in six, across 2014 to 2023. The 95% uncertainty interval was 12.8% to 20.1%.”
Link the report beside that wording, keeping fuller context immediately available if space is tight. Avoid switching “people” to “couples”, calling it a current local rate or presenting it as evidence that romantic relationships are failing. The estimate cannot establish those conclusions.
Before comparing it with another survey, check the age range, dates, question and recall period. A different percentage may reflect a different measurement rather than a real increase or decrease.
Let the headline open a question at home
Back on the sofa, your partner’s presence can matter and something can still be missing. Try, “I’m not using the number to judge us. I’ve missed having someone to talk with after work. Could I tell you about that?”
If a conversation is welcome, name one kind of connection you miss and ask what response is possible. It might involve time together, an old friend or wider community. A partner does not have to become your entire social world. Neither person owes touch, disclosure or agreement because a headline has been shared.
If loneliness is difficult to cope with, consider qualified support without waiting to match a percentage or duration. The NHS describes possible help. An app cannot replace appropriate professional care. Leave the number as context; make room for the person.
This adult educational guide explains population evidence and offers an original conversation example. It is not a diagnostic tool, individual health advice or a treatment.

