The numbers describe more than a gentle thinning of the contact list. In 1990, 55 per cent of American men said they had at least six close friends who were not relatives. By 2021, only 27 per cent did. At the other end of the scale, the share reporting no close friends rose from 3 per cent to 15 per cent.
The figures appear in the Survey Center on American Life’s 2021 State of American Friendship report. They compare answers from its May 2021 American Perspectives Survey with a Gallup poll conducted in January 1990.
This result comes from one historical comparison, not a continuous census of friendship. Both surveys were designed to represent American adults, but their methods were not identical. The broader signal is supported elsewhere; the precise size of the fall is less certain than the headline makes it look.
Two representative surveys, conducted in different ways
The underlying question was simple. Respondents were asked not to count relatives and to give their best estimate of how many close friends they had. Nobody supplied a definition of “close”, leaving each person to draw that boundary.
Gallup’s 1990 poll used telephone interviews with 1,226 adults. The 2021 survey included 2,019 adults from the Ipsos KnowledgePanel, a probability-based online panel. Interviews ran from 14 to 23 May, in English and Spanish, and the overall margin of error was plus or minus 2.4 percentage points. Estimates for men alone carry more uncertainty because they come from a subgroup.
The later survey found smaller networks among women as well. The share of women reporting six or more close friends fell from 41 per cent in 1990 to 24 per cent in 2021; 10 per cent of women in 2021 reported none. Men showed the steeper change, but this was not exclusively a male pattern.
The comparison has a meaningful method problem
A phone interview and an online questionnaire do not always produce interchangeable answers. Someone speaking to a live interviewer may hesitate to say they have no friends, or may round their number upwards to avoid embarrassment. The Survey Center acknowledges this possible social desirability effect in its own report.
The questions also arrived in different contexts. Sociologist Claude Fischer compared the original questionnaires and found that the 2021 friend-count question followed roughly three dozen prompts about social life, personal feelings and emotional support. Gallup’s sequence was shorter and had already asked respondents to estimate their friends more broadly.
By the time 2021 participants reached “close friends”, they had been encouraged to think carefully about disclosure, support and frequency of contact. That may have raised the threshold for whom they counted. The 2021 wording also began, “Now, thinking only about the friends you are close to”, adding another nudge towards a narrow circle.
The data may still capture a large social change. They simply cannot tell us how much of the 28-point fall came from people’s lives and how much came from the measurement.
Evidence beyond this comparison points the same way
Gallup asked the close-friend question again by telephone in 2001 and 2003. Its published results show stability through that period: 47 per cent of all adults reported six or more close friends in 1990, compared with 45 per cent in 2003. The most visible contraction seems to have arrived later.
In 2024, the Survey Center returned to the subject after pandemic restrictions had receded. Seventeen per cent of American adults reported no close friend outside their family, and only 25 per cent reported six or more. Whatever happened in 2021 was not followed by an obvious rebound.
Fischer’s wider review also found supporting evidence in behavioural measures. Time-use research indicates that Americans spent substantially fewer minutes with friends in 2019 than in 2003, with the decline accelerating during the 2010s. Another long-running survey recorded a fall in the share of young, unmarried adults who socialised with friends every week.
No single series settles the question, but the collection is harder to dismiss as a quirk of one questionnaire.
Friend count is not the same as friendship quality
There is no evidence-based rule saying six is the required number for a satisfying social life. Two dependable friends may offer more companionship than ten people who rarely speak. The upper half of the headline measures network size, not trust, support or contact.
Reporting zero is still a distinct answer. It means a person could not identify one non-relative who met their own standard of closeness. It does not establish that they were lonely, lacked meaningful family ties or wanted a larger social circle, but it marks the absence of one particular kind of relationship.
Scandinavia Standard has previously examined how adult life quietly removes the conditions that make friendship easy. People leave schools, teams and workplaces. They move, change shifts, care for children or parents, and lose the repeated unplanned contact from which closeness often grows. None of those changes requires someone to reject friendship deliberately.
More recent research complicates the gender story
Pew Research Center surveyed 6,204 American adults in September 2024 and found that men and women were about equally likely to have at least one close friend. Overall, 18 per cent said they had none outside their family.
There were gender differences in how those relationships were used. Some 54 per cent of women said they would be extremely or very likely to turn to a friend for emotional support, compared with 38 per cent of men. Women with close friends communicated more often by text, social media, phone and video. Men were slightly more likely to see a friend in person frequently.
That does not cancel the 1990-to-2021 result. It shows that “male friendship” contains several questions. How many friends someone names, how often they communicate, whether they meet in person and whom they trust with a difficult conversation do not necessarily rise and fall together.
The trend is clearer than its exact magnitude
The 55-to-27 and 3-to-15 comparisons deserve attention, but not false precision. The surveys changed mode, question order and conversational context over 31 years. The word “close” remained subjective throughout.
A restrained reading still leaves a substantial finding. Multiple surveys now show many Americans reporting very small friendship networks, while time-use data suggest less time spent with friends, especially from the 2010s onward. Something appears to have contracted. The exact amount cannot be recovered from two percentages alone.