Flirting or Just Being Friendly? What the Research Can Actually Tell You

Flirting is ambiguous by design, and reading harder will not fix that

Start with what flirting is for, because it explains why the question is so hard to answer.

David Henningsen and Mary Lynn Miller Henningsen, summarising the research programme on flirting motives, define flirting as goal-driven behaviour that promotes ambiguity about the flirter's motivations. The ambiguity is not a side effect. It is part of the definition.

Steven Pinker, Martin Nowak and James Lee gave the general reason in 2008. Their paper on the logic of indirect speech takes sexual come-ons as one of its central examples, alongside veiled threats, polite requests and concealed bribes. An indirect overture allows plausible deniability: a willing listener can take it up, and an unwilling one has nothing explicit to react against. Lee and Pinker extended the theory in 2010 with a point that matters here. Even when both people can work out what was meant, saying it indirectly still stops it from becoming common knowledge — the thing each person knows the other knows. That is what lets a friendship carry on after an overture that went nowhere.

Two cautions about that source. It is theory, supported by a game-theoretic model and by experiments in which participants judged the intentions of characters in written scenarios, not by recordings of real flirting. And it describes a function rather than a guarantee: plenty of ambiguity is simply two people who do not yet know what they want.

The consequence holds either way. A signal built to be deniable cannot be decoded with confidence from the outside, because resisting that is part of its job. It is why confident lists of tells keep failing, and why the measured hit rates are so low. In the study that recorded strangers talking and then asked both of them, flirting was detected in 7 of the 25 cases where it happened — 28% — while its absence was recognised 84% of the time. The full numbers and their caveats are in what the evidence supports about reading someone's interest, and the other pages that try to read interest from the outside are collected under Signs.

There is a second cost to the ambiguity, and it lands on the person doing the reading. Not knowing whether someone likes you can itself make you think about them more, which is easy to mistake for evidence that something is there; the one experiment on that is described, with its limits, in why smart people keep choosing the wrong partners.

People flirt for reasons that have nothing to do with wanting you

The question "flirting or friendly?" assumes that flirting means interest. The research on why people flirt says that often it does not.

Henningsen's 2004 study in the journal Sex Roles set out six motives drawn from the earlier literature: sexual; relational, developing or deepening a relationship; exploring, finding out whether a connection is possible at all; fun; esteem, feeling better about oneself; and instrumental, getting something from the other person. His later work sorts them into two families. Sex, relational and exploring are courtship motives. Fun, esteem and instrumental are quasi-courtship — the courtship repertoire used for something other than courtship.

The sex differences in that study explain a great deal of everyday misunderstanding. Men tended to see flirting interactions as more sexual than women did; women attributed more relational and fun motives to them. There were no sex differences for esteem, exploring or instrumental motives. Two people can be in the same exchange and be describing different events.

Henningsen, Braz and Davies then compared working adults with students in 2008. The working sample was less likely than the students to see flirting as sexually or relationally motivated, men reported more sexual and exploring motives than women, and instrumental flirting turned up more at work than in social settings.

None of that is a cynical reading of other people. Most of those motives are harmless, and flirting for fun is one of the pleasures of social life. It is a reason not to treat the presence of flirting as settling the question of interest. What flirting settles is smaller: that the person is, for some reason, playing.

The limits belong here too. These are self-report studies of how people understand flirting in general, not observations of particular people, and the samples are largely American students and workers.

Style decides how interest looks, and one style hides it

A second reason the question resists an answer is that people differ widely in how they show interest in the first place.

Jeffrey Hall, Steve Carter, Michael Cody and Julie Albright developed the Flirting Styles Inventory on a sample of 5,020 adults and found five distinct styles: traditional, physical, sincere, playful and polite. Women scored higher than men on every style except playful. The physical, sincere and playful styles went with more dating success, and the physical and sincere styles with important relationships that escalated faster and had more emotional connection and physical chemistry.

The follow-up is the one that matters for this question. Hall and Chong Xing recorded 51 pairs of opposite-sex strangers talking for ten to twelve minutes, had four independent coders score 36 verbal and nonverbal behaviours, and looked at how people with each style behaved when they were physically attracted to their partner. The behaviour matched the style. Playful flirts became more obviously engaging and flirtatious. Sincere flirts smiled more, and fidgeted, teased and got distracted less. Polite flirts, when attracted, became more reserved and more distancing.

Put those findings side by side and the difficulty is plain. The most visibly flirtatious person in the room may be a playful flirt, for whom that register is a habit of style before it is a message about anyone. The person who seems a little cooler with you than with everyone else may be a polite flirt who is interested. Visible intensity is a measure of style far more than of feeling, which is why there is no single signature of interest to watch for.

Hall's samples were heterosexual, and the recorded conversations were between strangers in a laboratory. How the styles play out between long-standing friends, or between people of the same sex, has not been measured in the same way.

The error almost nobody warns you about

Most writing on this subject worries about one mistake: taking friendliness for interest. The study that measured both mistakes at once found that the opposite one was bigger.

Coreen Farris, Teresa Treat, Richard Viken and Richard McFall showed 280 heterosexual undergraduates a set of 280 full-body photographs of women, for either half a second or three seconds each, and asked them to sort every photograph into one of four categories: friendly, sexually interested, sad or rejecting. The photographs had been chosen so that most men and most women on an earlier panel agreed on the category, and attractiveness was held constant across the four sets.

The friendly photographs were sorted correctly most of the time — 79.9% by men and 87.0% by women — and read as sexual interest 12.1% of the time by men and 8.7% by women. That is the familiar error, and it replicated.

The photographs showing sexual interest were a different story. Men sorted them correctly 48.9% of the time and women 59.2%. Men read them as merely friendly 37.8% of the time, and women 31.9%.

Two things follow. First, for both sexes the largest confusion in the whole task was reading interest as friendliness, not the reverse, and both sexes were better at recognising friendliness than at recognising interest. If you have been assuming the risk runs only one way — that anything warm is probably just warmth — this is the best measurement available saying you are more likely to be missing interest than imagining it.

Second, the sex difference, which reframes an old argument. The standard explanation for men seeing more sexual intent is a lower threshold: men supposedly label ambiguous cues as sexual more readily. Farris and colleagues used a model that separates threshold from sensitivity, and found no evidence that men had the more lenient threshold. What differed was sensitivity. Men were worse at telling friendliness and sexual interest apart, so they made more errors in both directions.

The limits are real and go in the same breath. These were photographs of women, judged in seconds, and "correct" means agreeing with the earlier panel rather than matching what the women photographed actually felt. It is a laboratory measure of perception, not a record of anyone's evening. Whether men also overestimate women's intent in live interaction is a separate debate, and the challenge to the overperception finding, and the preregistered test that answered it, are laid out elsewhere on this site.

Who misreads friendliness most, and why it is not a harmless puzzle

When the error runs the other way — ordinary friendliness read as sexual interest — it is usually minor and quickly corrected. Sometimes it is not, and this is where the question stops being a parlour game.

Antonia Abbey's first study, in 1982, had 144 undergraduates hold or watch five-minute conversations, and found men rating the woman in the conversation as more seductive and promiscuous than women did. Her 1987 follow-up surveyed undergraduates about incidents in their own lives. A large proportion of both women and men had had their friendliness taken as sexual interest, more women than men; Farris and colleagues, summarising it, put the figure for women at 67%. Most of those incidents were resolved quickly and without trouble. Some involved forced sexual activity and left the person angry, humiliated and depressed.

In 2007 Angela Jacques-Tiura, Abbey, Michele Parkhill and Tina Zawacki asked who makes the error most often. Among 356 male college students, how often men reported misperceiving women's sexual intent was predicted by hostile masculinity, an orientation towards impersonal sex, and drinking in dating and sexual situations. The more of those risk factors a man had, the more often he reported misperceiving. The measures are self-reports and the design is a snapshot, so it cannot show what causes what. What it does show is that the error is not scattered evenly across people. It clusters with attitudes.

Lindgren, Parkhill, George and Hendershot's 2008 review describes the basic sex difference as replicated in scores of studies using a variety of methods, and links it to sexual assault and harassment. That is why this site does not treat "mixed signals" as a neutral topic. A reading of someone's friendliness is a guess about another person. It is never permission for anything, which is the subject of a section further down.

The one pattern with evidence behind it: is the warmth aimed at you?

Now the useful finding. The previous version of this page said that flirting is directional and friendliness is distributed, and it said so without evidence. The evidence exists, and it is more interesting than the claim was.

Paul Eastwick, Eli Finkel, Daniel Mochon and Dan Ariely ran seven speed-dating sessions with 156 undergraduates, each of whom had four-minute dates with nine to thirteen people and rated their romantic desire for each one. Because everyone rated many people and was rated by many, the researchers could separate two things: how much a person desired their dates in general, and how much they desired one particular person over and above that.

The two worked in opposite directions. Unique desire was reciprocated: when someone desired a particular partner more than their other dates, that partner tended to desire them more in return (r = .14). General desire was not. People who desired all of their dates were desired less by those dates (r = −.41), and part of that effect ran through how unselective they seemed — their partners' guesses about how many of the other people there this person would say yes to.

The contrast with friendship is the striking part. In non-romantic settings, as Eastwick and Finkel note in their review of the method, people who like others in general tend to be liked back, at an average of about r = .43 among people meeting for the first time. Liking everyone makes you likeable. Desiring everyone does not.

So selectivity is real, and partners picked up on its absence within four minutes. It is also exactly the thing a list of behaviours cannot capture, because it is not a behaviour. It is a comparison. To see it you have to watch how the same person treats other people in comparable situations — which is why a single warm evening, however warm, tells you so little.

The limits, stated plainly. These are ratings of desire after four-minute dates, not observed behaviour among friends or colleagues, and the reciprocity correlation, though reliable, is small. Speed dating also makes romance the explicit frame, which an office or a friend group does not. What transfers is the logic, not the numbers. And it is not a prediction engine: in the same paradigm, whether two particular people click could not be predicted from anything either reported in advance, and between people who already know each other, attraction develops by a different route.

Friends flirt, and stay friends

The question most often comes up about a friend rather than a stranger, and the research on friendship adds two complications.

The first is that flirtation is ordinary inside friendships that never become anything else. Kori Egland, Brian Spitzberg and Michelle Zormeier asked people about cross-sex platonic and romantic relationships and identified four kinds of flirtation behaviour: display, stereotyped, attentiveness and conversational. People in both kinds of relationship reported flirting, to varying degrees. What differed was how the flirtation was displayed and how it related to judgments of appropriateness and communication competence. Flirting between friends is not a malfunction or a confession. For some friendships it is simply part of the texture — and it carries on inside romantic relationships too, where what actually keeps couples going has its own evidence.

The second complication is projection, and it may be the most important finding on this page for anyone trying to read a friend. Bryan Koenig, Lee Kirkpatrick and Timothy Ketelaar ran two online studies with members of opposite-sex friendship pairs — 238 people in the first and 198 in the second — asking each person about their own sexual and romantic interest in the friend and about how interested they thought the friend was in them. People projected: their own interest shaped what they believed the friend felt. And that projection accounted for the familiar sex difference. Men overperceived and women underperceived their friend's sexual interest — though not their romantic interest — because each was reading their own feelings into the other person.

Bleske-Rechek and colleagues found the same shape with pairs of friends, with larger numbers attached: each friend's estimate of the other's attraction tracked their own attraction far more closely than it tracked what the other person reported. The figures are in the research on why people stay in the friend zone without ever being turned down.

The practical meaning is uncomfortable and simple. When you are interested in someone, your reading of their interest is partly a readout of yours. The more you want it to be flirting, the less your impression that it is flirting is worth.

Teasing and laughter: what they indicate, and what they do not

The previous version of this page leaned on teasing, and on a claim about John Gottman's research on humour in couples that it never tied to a paper. That claim is gone. There is better evidence, and it points somewhere more interesting.

Norman Li, Vladas Griskevicius and colleagues tested two explanations of why humour matters in attraction. One says being funny advertises good qualities. The other says humour works as an interest indicator — a way of starting relationships and checking on existing ones. Across three studies, both men and women were more likely to start humour, to respond positively to the other person's humour, and to rate the other person as funny when they were initially attracted to them.

Read that carefully, because it cuts in two directions. It supports the idea that someone who keeps joking with you and laughs readily at your jokes may be interested. It also means that finding someone funny is partly a symptom of your own attraction, which is the projection problem again. And the authors frame humour as a signal of interest in social relationships generally, new and existing, not only romantic ones. Friends laugh at each other's jokes for the same reason.

Teasing is more complicated than humour. Dacher Keltner and colleagues' review defines a tease as a playful provocation in which one person comments on something relevant to the target, marked as play by what they call off-record markers — exaggeration, metaphor, intonation, facial expression, laughter. People tease to socialise, to flirt, to resolve conflicts and to pass the time, and with slight changes in delivery the same act humiliates or harasses. The review found teasing more frequent, and more hostile, when it came from higher-status and familiar people and from men, with sex differences smaller than usually assumed.

So teasing tells you that someone is comfortable enough to provoke you. It does not tell you why. And teasing that lands on something you are sensitive about, continues after you have stopped laughing, or comes from someone with power over you is not a flirtation signal waiting to be decoded. It is the other outcome Keltner describes.

Remembering what you said is attention, and friends pay it too

The previous version of this page claimed that "memory is downstream of caring", and treated someone remembering a small detail from weeks ago as a romantic tell. We could find no study that tests remembered details as a sign of romantic rather than platonic interest.

What does exist points the other way. Remembering and following up are part of what relationship researchers call responsiveness — the other person's sense of being understood, accepted and cared for — and responsiveness is the machinery of closeness in every kind of relationship, as set out in how people come to open up to each other. A good friend remembers. So does a good manager. So does someone who is interested in you.

This is the same wall every list of signs runs into. The largest meta-analysis of attraction and behaviour, by Montoya, Kershaw and Prosser, pooled 309 effect sizes and found eleven behaviours reliably related to attraction — gaze, smiling, laughter, mimicry, nodding, closeness, talking — with an overall association of z = .20. Montoya's own reading, in the University of Dayton's summary, is that these are the behaviours people use to build trust and rapport, and that the same set turns up between doctors and patients, parents and children, and salespeople and customers.

Attention is worth noticing. It is good evidence that someone values you, and weak evidence about the kind of value.

At work, the same behaviour carries different costs

The workplace needs its own section, because power and consequences change what flirting does.

Henningsen and colleagues' 2008 comparison found instrumental flirting — flirting to get something — more common at work than in social settings. In 2023, Henningsen and Henningsen put 120 people, 60 women and 60 men, into four-person mixed-sex groups making decisions, and measured how members saw each other. The more someone was seen as using charm, the less task-competent the others judged them to be. It is one study, with students in short-lived groups, but it fits the wider point: at work, flirting is readily read as a tactic, and it costs the person doing it.

For the person on the receiving end, the question at work is not only whether it is flirting but whether it is welcome. Under US federal law, harassment is unwelcome conduct based on a protected characteristic, sex included; isolated minor incidents are generally not unlawful, while conduct that is severe or pervasive, or that is made a condition of keeping the job, can be. The Equal Employment Opportunity Commission encourages employees to tell the person directly that the conduct is unwelcome and must stop, and to report it early. Other countries have their own frameworks, and an employer's policy may go further than the law.

And if you are the one wondering whether a colleague is flirting — especially a colleague who is junior to you — the costs are lopsided. A person who reports to you has fewer safe ways to decline, and almost everything on a list of "signs" is also what people do to keep a working relationship smooth with someone who has power over them.

A reading is never permission

This is the section the rest of the page has been building towards.

Everything above says that the behaviours of friendliness and flirtation overlap almost completely, that people misread them in both directions, that some people misread them far more than others, and that your own desire bends the reading. Any one of those would be a reason not to act on a reading as though it were an answer. Together they settle it.

Research on consent shows where the gap opens. Kristen Jozkowski and colleagues asked 185 US college students, in open-ended questions, how they define, communicate and interpret sexual consent. Men and women did not differ in how they defined it. They differed in how they signalled it — women reported more verbal strategies, men more nonverbal ones — and in how they read a partner, with men relying more on nonverbal cues. The authors suggest that difference may help explain some misunderstandings. Nonverbal cues are exactly the channel flirting uses, and exactly the channel built to be deniable.

So the line is simple, and it does not depend on what the signals were. Flirting, friendliness, a long conversation, laughter, a drink accepted or a kiss earlier in the evening is not agreement to anything beyond itself. Agreement to something is given for that thing, can be withdrawn, and is established the plain way: by asking and listening to the answer. Someone who pushes past a no, or treats hesitation as a yes because of how a person behaved earlier, is not misreading a signal. That is coercion.

If it has happened to you, it was not caused by how friendly you were. Find A Helpline (findahelpline.com) lists free, confidential services by country, and in the United States RAINN's National Sexual Assault Hotline is 1-800-656-4673, with online chat at rainn.org.

What actually resolves it: a question that is easy to say no to

If the signals cannot settle it, language can — and asking costs less than it feels like it will, with one catch.

Samantha Joel, Rimma Teper and Geoff MacDonald found in two studies that single people were significantly less willing to turn down an unsuitable potential date when they believed the situation was real than when it was hypothetical, and that the gap was partly explained by not wanting to hurt the other person's feelings. The rejection you are bracing for is usually gentler than the one you picture. The catch is that the same concern produces soft yeses and vague maybes that mean no.

That gives the shape of a good question. Ask once, plainly, about something specific — a coffee, a film, a walk on Saturday — rather than making a declaration the other person has to answer as a verdict on the whole friendship. Make declining easy, because only a question that is easy to decline produces an answer you can trust. Take the first answer at face value: a warm "things are really busy right now" with no alternative offered is an answer. And do not run the question again in different words to see whether it comes out differently.

What to avoid is what this genre usually recommends instead: escalation disguised as something else, engineered so the other person has to guess what you meant. That hands the ambiguity back to them, and it tells you very little. The same goes for strategic waiting; when the timing rule was actually tested, the delay mostly made people seem less interested and less reliable, and numbered rules in general have no research behind them at all.

If the answer is yes, the reading is over and a different question starts, which the signs a new relationship has real potential is a better guide to than any cue, beginning with how the first date tends to go wrong. If the answer is neither yes nor no for months on end, that is its own pattern, and it has a name: the situationship. It is not a rare one: 39% of US adults in one national survey said they had been in one, a lifetime figure rather than a count of people in one now.

What this page used to say

This article was rewritten on 16 September 2026. The previous version was a list of nine "signals", written in the first person, and it rested on five references that supported none of its specific sentences: a 1971 book by Albert Mehrabian, Eibl-Eibesfeldt on the eyebrow flash, Hess on pupil size, Hazan and Shaver on adult attachment, and a popular book by Helen Fisher. None of them is about telling flirting from friendliness. The eyebrow flash is a greeting, the pupil tracks arousal of any kind, and Mehrabian's figures are the most misquoted numbers in this field — the 93% figure is taken apart here.

Five specific claims have been removed. "Memory is downstream of caring", offered as a romantic tell: no study supports it, and the section above explains what remembering does indicate. A test based on how someone behaves when another attractive person arrives, which described the newcomer as "a mild competitive variable": there was no evidence for it, and it invited reading possessiveness as interest — the confusion taken apart in what the research says about jealousy.

"Whether they make it easy for you to leave", presented as diagnostic, had no source. A description of Gottman's research on teasing in couples was never tied to a paper, and was followed by a note about a failed replication that concerned a different Gottman finding; it is replaced by Li and colleagues (2009) and Keltner and colleagues (2001). And an FAQ answer said that proximity shifts and small grooming gestures are partly involuntary signs of attraction. Primping showed no relation to liking in the largest meta-analysis; proximity did, and it is just as much a sign of friendship.

One idea survives, because it turned out to have evidence the old page never cited: that interest shows up as selectivity rather than intensity. It now rests on Eastwick and colleagues (2007).