Friendship, conflict and boundaries

Autistic people report more loneliness than non-autistic people, on average. Pooling 23 studies of children and adults, a meta-analysis found a large difference.¹ In one Australian study, 220 autistic adults reported more loneliness than 146 non-autistic adults; they named noise, difficulties with communication and past bad experiences as barriers to socializing.² Loneliness goes with depression and, less strongly, with anxiety; the pooled studies all measured both at one time, so which comes first is unknown.¹ In one survey of 108 autistic adults, those with more friends and better friendships were less lonely, and number of friends went with self-esteem, depression and anxiety over and above loneliness.³

A review of 34 studies of autistic adults found that what loneliness means for them is only partly measured. Not fitting into society, which the authors call collective loneliness, appears to matter to autistic adults but was rarely measured, and only one study used a scale written for autistic adults.⁴

Loneliness is not the same for everyone. In one small US group followed from about age 9 to 25, about a third of whom had other developmental conditions rather than autism, people became lonelier from adolescence into adulthood on interview ratings, though a questionnaire showed no clear change. A sizable minority said they were not lonely, and those who coped by distracting themselves were lonelier.⁵ In a survey of 209 adults in the same study, autistic adults reported more loneliness, and loneliness statistically accounted for part of the link between sensory sensitivity and anxiety and depression; in interviews, nine autistic adults described the pain of loneliness alongside a need for solitude to recover from social overstimulation.⁶ A review of 14 correlational studies, mostly of children and mostly male, found that lower interest in socializing went with more anxiety and depression, though each pooled estimate rested on only three to five studies and the authors call the results preliminary.⁷

A scoping review of 22 studies, across ages, found that autistic people define friendship much as other people do, by similarity and nearness. The challenges it found were more specific: insecurity about friendships, a narrow and deep focus of attention, and efforts to conform to non-autistic norms that led to anxiety.⁸ Young autistic adults in the US said friendships worked best where their differences were treated as normal, through accepting friends and shared interests; those who were isolated did not know how to find a group.⁹ Forty autistic adults described taking part socially in five settings: work, neighborhoods, interest groups, support services and inclusive settings, and online.¹⁰ In an online survey, autistic people valued the greater control over and understanding of communication that the internet gives, and enjoyed meeting people online more than non-autistic respondents did.¹¹

A lower score on a general friendship measure does not always mean less closeness. In an online survey of 949 adults, 532 of them autistic by their own report and with groups not matched on age, autistic adults scored lower on a general friendship questionnaire but rated their closest relationship as closer than non-autistic adults did. More autistic respondents, though, could not name a close relationship at all.¹²

In joint interviews with 11 pairs of friends, 20 autistic young adults and their autistic or non-autistic friends spoke of acceptance, authenticity, flexible give and take, and depth rather than how often they met.¹³ Friendships with other autistic people are often described as easier (see Unmasking with others), but in written answers from 142 autistic adults in Australia and New Zealand, these relationships were not always smooth: some had conflicts over differences in compatibility, and a shared neurotype on its own did not guarantee getting on.¹⁴

One program for young autistic adults is PEERS, a group course that teaches rules for making and keeping friends. Two small randomized trials by its developers at UCLA, in young adults without intellectual disability, compared it with a waiting list; the second had 22 participants (12 and 10). Both involved caregivers in the course. They reported gains: less self-reported loneliness and more get-togethers reported by caregivers in the first, and better social skills and engagement in the second, mostly still there 16 weeks later.^16,17 A later study adapted PEERS for a college campus, with six autistic students and five non-autistic partners and no comparison group; the students reported better social reciprocity afterwards, and two of the six showed reliable improvement on social-skills knowledge, social responsiveness or quality of socialization.⁴²

Pooled estimates for adults come mostly from trials that compare a program with a waiting list or usual care. A preregistered meta-analysis pooled 35 studies of all kinds of programs for autistic adults, not only social-skills courses (1,631 adults): it found small to moderate gains in mental health and social cognition (effect sizes of 0.35 and 0.45) and a larger gain in quality of life (0.64), but the social-cognition and quality-of-life results varied widely between studies (I² of 68% and 80%, where a higher number means more disagreement), effects on behavior, thinking skills, daily living and employment were not significant, and its findings are not specific to friendship programs.⁴³ An earlier meta-analysis of five trials of group social-skills training (145 adults) found large effects on parent-rated social responsiveness and warned they may be inflated; none of those five trials compared the training with an active treatment.¹⁸

In children, a double-masked trial assigned 84 autistic children aged 8 to 12 in Australia to KONTAKT or to a group art program, an active comparison; both groups improved on their personal social goals, and friendship quality was the only outcome scale that favored KONTAKT, though the children enjoyed the KONTAKT sessions more.⁴⁴

Some researchers and autistic adults question the approach. A commentary argues that teaching social rules can make interaction less authentic and add to stigma.¹⁹ Eleven UK autistic adults asked for support tailored to them; some had bad experiences of social-skills training, and some said non-autistic people should be the ones trained.²⁰

Research on conflict in autistic friendships is thin. Among 102 UK adolescents, the 27 autistic girls reported more conflict with friends than the 26 non-autistic girls and found it harder to manage, while autistic boys' friendship patterns differed from all the other groups. The kind of conflict differed by gender regardless of diagnosis, and the authors conclude that gender, more than diagnosis, shaped these experiences.²¹ In a mixed-methods study of 19 autistic and 19 non-autistic women, relationships were broadly similar, and the autistic women described themselves as happier and more self-assured in adult relationships than they remembered being as adolescents. They also described more social and sexual vulnerability, which they put down to difficulty reading social situations.²²

Some work looks at how autistic people keep an interaction on track with each other. Filmed playing video games together, 30 young autistic adults from a charity (24 diagnosed, 6 awaiting assessment) built shared understanding by generously assuming common ground; most were men and all spoke fluently. When the others followed, rapport came quickly; when they did not, the remark could disrupt the game. A low demand for coordination, asking little of each other in keeping the game in step, absorbed many of those disruptions.²³ Sharing a neurotype does not remove conflict, though: autistic adults also report falling out with autistic friends.¹⁴

Whether autistic adults find it harder to say no has been measured mainly as compliance: going along with requests to please or avoid conflict. Two small UK studies with an author in common, each of about 26 autistic adults, found them more compliant than comparison adults on a questionnaire. One also found it in parents' reports, the other on a behavioral task.^24,25 In that second study, across both groups, self-esteem was the only measure that independently predicted questionnaire compliance (bullying and anxiety correlated with it but added nothing), and questionnaire scores did not predict behavior on the task.²⁵ A third study, which recruited 32 autistic and 30 comparison adults, mostly men, found no difference.²⁶ Compliance differs from suggestibility, being swayed by leading questions. A meta-analysis that combined suggestibility and false memory into one measure found autistic samples, taken together, less susceptible than comparison groups.²⁷

People's own accounts describe the difficulty more plainly. In written answers to an online study, 43 autistic adults recruited through social media described repeated victimization by friends and acquaintances, difficulty saying no and difficulty spotting manipulative intent.²⁸ Nine late-diagnosed Spanish women and non-binary people described crossing their own boundaries to belong and keeping relationships at all costs.²⁹ No study found as of October 2026 has measured assertiveness or saying no directly in autistic adults.

Autistic adults report being bullied, exploited and assaulted more often than non-autistic adults. In an online survey run from Cambridge, UK, of 426 autistic and 268 comparison adults, 48% of the autistic group (against 20%) had been tricked or pressured into giving money or possessions, 70% (against 31%) had been bullied by someone they considered a friend, and, of those who answered about partners, 20% (against 9%) had been forced into sexual activity by a partner.³⁰ Among US college students, past-year emotional aggression was 44% for 1,411 autistic students against 26% for non-autistic students.³¹ In an international online sample, mostly women and mostly in Australia, 75.4% of 118 autistic adults had been sexually harassed since age 15 and 58.5% stalked, against 56.4% and 27.3% of 110 non-autistic adults.³² In Canada, 45 autistic adults reported more victimization in childhood and, as adults, more teasing and unwanted sexual contact than 42 non-autistic adults, with no difference in harming others.³³

It often starts early. A meta-analysis of 17 studies of autistic young people found that 44% had been bullied at school, more than their typically developing peers.³⁴ Studies that account for other conditions give a less clear picture of how much of this belongs to autism itself (see the next section).

The common explanation is that autistic people are too trusting or cannot read deceit. One online experiment, using written scenarios, does not support the second part. In it, 72 autistic adults spotted suspicious behavior as well as 64 non-autistic adults, despite lower scores on mind-reading tasks. The authors note that the idea that mind-reading difficulties draw autistic people into crime has "virtually non-existent" scientific support, though across both groups weaker mind-reading did go with poorer detection.³⁵ In a study by the same group, 1,410 viewers rated truthful autistic adults as more deceptive and less credible than non-autistic adults, which may make it harder to be believed.³⁶ The small Canadian study found no evidence that social-communication ability or emotion regulation explained the higher risk, though it was too small to rule this out.³³

Large studies of young people suggest the excess risk is not specific to autism. In Swedish registers of 1.34 million people followed from age 15, autism went with more violent victimization serious enough to need hospital care, but only in females; after comparing siblings and adjusting for externalizing problems, only ADHD remained.³⁷ In about 4,500 Swedish twins, autism traits rated by parents in childhood carried almost three times the risk of coercive sexual victimization by age 18 for girls (the estimate for boys was similar but not statistically significant), and the difference disappeared once overall neurodevelopmental traits were taken into account, a link the authors trace to shared genes.³⁸ Among 270 autistic and 270 non-autistic US college students matched on background and other diagnoses, autistic students were no more likely to report unwanted sexual contact or assault; ADHD predicted both.³⁹

In interviews, 22 autistic adults who had experienced violence named being "too trusting" and going along with people to please them as risks, and said pressure to fit in had shaped both; the violence had damaged their trust.⁴⁰ Many of 102 autistic people interviewed about being hurt by people they knew described it as normal and masked to stay safe, which exhausted them; some felt they had deserved it.⁴¹

Many autistic writers say the hard part is finding friends, not being able to have them. Daniel M. Jones writes: "It is not that, as autistic people, we cannot make friends; our difficulty lies in finding friends."^c1 Sol Smith describes autistic people building friendships around shared interests, "such as activities, hobbies, or pop culture", and suggests interest groups such as trivia nights or Dungeons and Dragons.^c2 Megan Griffith writes that autistic people "tend to make friends with other neurodivergent folks, or with extremely compassionate people who are happy to make room for our autism". Griffith also names friendship trauma, "traumatic past events involving friends", and says nearly every autistic, ADHD and AuDHD person Griffith knows has some.^c3

Some writers say friendship with other autistic people matters in its own right. Jim Sinclair, recalling the early years of Autism Network International, quotes Katherine, a woman diagnosed in 1993, who wrote after first meeting other autistic people: "I believe all persons with Autism need the opportunity to become friends with other Autistic people."^c11 Jean Winegardner, diagnosed at 38, describes having built "a neurotypical community" that became harder to be at peace with, and names autistic community as the one thing that could have made life easier since childhood.^c12 Lydia X. Z. Brown rejects the clinical account of autistic loneliness as a lack of social skills, recalling being "terrified of making my friends upset or angry at me".^c13 Pete Wharmby adds a caution: shared autism does not mean "all autistic people will get on with all other autistic people".^c14

On conflict, writers describe both avoiding it and learning to repair it. Megan Anna Neff writes that conflicts "can easily overwhelm Autistic people due to our heightened sensory and nervous systems, resulting in conflict avoidance".^c4 Griffith describes rupture and repair: every relationship has ruptures, and both people then work to close the gap, and Griffith suggests therapy or coaching as a safe place to practice it.^c5 Sarah Nannery offers "repairing strategies" for when a conversation goes wrong, such as matching the other person's emotional level and stating what you see: "I see that this is important to you. It's important to me, too. But I don't know what to do about it."^c6

Writers tie difficulty saying no to masking. Smith calls people-pleasing "the highest form of masking".^c7 Devon Price describes fawning, compulsive people-pleasing, as attractive because it "flattens social interactions into one easy rule: always say 'yes'".^c8 In a later book Price defines a boundary as "an expectation you set for how others should treat you, combined with a consequence you will personally enact if someone's conduct does not meet it", and suggests starting with small no's, such as asking to open a car window.^c9

On exploitation, Neff writes that "many Autistic people are inherently trusting, coupled with a tendency to people-please", which leaves them open to manipulation, and Nannery warns that taking people at their word can make an autistic person "more vulnerable to manipulation".^c10,c6 Wharmby draws a line between trust and gullibility when describing mate crime, abuse by people believed to be friends: "Autistic people are prone to this not because we’re gullible, but because we’re often quite trusting, especially when people are kind to us."^c14 Daniel Burford warns that boundary advice written for friendships between equals does not carry over to a boss, a parent or a partner with power over you: "Setting a boundary and leaving safely are two separate tasks."^c15 The research above supports the higher rates of victimization, but in one experiment autistic adults spotted suspicious behavior as well as others.

Nonspeaking writers describe friends who also type or spell as a reason to communicate at all. Rishan Pavan, a new typer who was nine when he wrote, says he saw no point in communicating before he met his "autism brothers": "Now that I have friends just like me I see why I need my words to be heard."^c16 Kaylie Clinton, who spells on a letterboard, recalls practice meetings with a family friend's son where at first she could only say yes or no, "but it still felt like I finally had a real friend". Clinton calls it the most common misconception about autism that autistic people do not want "a connection with other humans".^c17