How to Flirt: A Practical Guide to Showing Romantic Interest
Flirting is communicating your own romantic interest clearly — not decoding someone else's. A research-grounded framework for warmth, playfulness, and honest directness.
Quick answer
Flirting means communicating your own romantic interest — through warmth, attention, playfulness, and honest words — not decoding someone else's. Research shows people often misread interest from cues, so the honest approach is: express your interest clearly, watch whether they reciprocate, and respond to what they do. Stop at any no.
How to read the labels in this guide. Research says = a peer-reviewed finding, with the study's actual dependent variable named. Editorial = our synthesis into practice. Analogy = a finding borrowed from a neighboring field, not flirting research. The chain from finding to advice stays visible.
What flirting actually is
Research says: flirting is defined by the flirter's intent, not by the behavior itself. Henningsen (2004) defines flirting as goal-directed behavior that attempts to signal or elicit sexual interest — which makes it a sender-side construct. That study also catalogued six self-reported motives (sex, fun, exploration, intimacy, self-esteem, instrumental goals); miscommunication follows naturally, because the receiver routinely attributes a different motive than the sender had. (Dependent variable: self-reported motives and meaning attributions — not interest, not outcomes.)
Research says: people communicate romantic interest in different styles. Hall, Carter, Cody, and Albright (2010) developed the Flirting Styles Inventory from 5,105 eHarmony users, replicated in a 4,500-person international survey: physical, traditional, polite, sincere, and playful — most people blend them with one dominant. (Dependent variable: self-reported communication style.) Caution: links between styles and "dating success" are retrospective self-reports — nothing here shows adopting a style causes better outcomes.
Research says: the same behavior can mean opposite things in different people. Hall and Xing (2015) found polite flirts leaned back and became less overt the more attracted they were. (Dependent variable: observed flirting behaviors.) Load-bearing for everything below: behavior has no fixed meaning.
Editorial: treat flirting as communication, not code. Your job is to express your own interest clearly — and to remember the other person's behavior is genuinely ambiguous.
Friendly vs. flirty: the line is blurrier than you think
Research says: friendliness is routinely misread as romantic or sexual interest. In Abbey's (1982) lab paradigm, male raters judged the same actor's friendliness as more seductive and promiscuous than female raters did — a pattern confirmed across 28 studies in a meta-analysis (La France et al., 2009). Neither study checked judgments against the actor's actual intent (dependent variable: perceived sexualization — rater differences, not measured errors).
Research says: context changes the meaning of the same behavior. Henningsen, Braz, and Davies (2008) found flirting in social settings was attributed to courtship motives, while flirting at work was attributed to esteem, instrumental, or fun motives. (Dependent variable: self-reported motives.)
Editorial: assume ambiguity, not interest. A warm, funny, attentive person may simply be warm, funny, and attentive — nothing in the literature licenses treating friendliness as attraction. For the full treatment of reading interest — and why the research says you'll often get it wrong — see Reading Body Language.
The governing principle: you can communicate your own interest. You cannot read their mind.
This is the site's position, and the research on interest perception is the reason for it.
Research says: people detect the absence of flirting far better than its presence. In Hall (2015), student pairs were highly accurate at knowing when their partner was not flirting, but only a minority detected actual flirting — judged against the partner's self-report. Outside observers did little better. (Dependent variable: detection accuracy.) If someone isn't showing interest, that's relatively legible. If they are, you probably can't tell from cues alone.
Research says: your own feelings distort your read. Speed-dating studies find that people perceive in others roughly the interest they themselves feel — projection of one's own sexual interest onto partners (Lee et al., 2020, with 1,226 speed-daters; Koenig et al., 2023; dependent variable: perceived vs. self-reported sexual interest). In one smaller study, men detected a partner's attraction more accurately when they themselves were not interested (Samara, Roth, & Kret, 2021, Archives of Sexual Behavior; 67 speed-daters, 277 dates; dependent variable: attraction detection accuracy). Lee et al. found the familiar sex difference in overperception shrank to negligible once sociosexuality and projection were accounted for — not a story about one sex being "wired" a certain way.
Editorial: your "read" is partly a self-report in disguise — be most skeptical of it exactly when you like someone most.
Research says: flirtatious behavior does not equal interest. In Place et al. (2009), observers judging recorded speed dates couldn't read the targets who were flirtatious but not actually interested — nobody could. (Dependent variable: judged vs. reported romantic interest.) Flirtation is deniable and ambiguous by design.
Research says: even facial expressions don't support confident individual inference. Barrett et al.'s (2019) review of 1,000+ findings concluded it is not possible to confidently infer emotions from facial movements — the same configuration expresses different states by context. Attraction has even less of a distinctive signature, so the limitation applies fully to "reading" romantic interest from faces. (Dependent variable: emotion inference — cited as an analogy, not flirting evidence.)
Research says (the honest counterweight): perception isn't pure noise — Eastwick et al. (2022) found modest genuine tracking of a match's actual desire. Some signal exists; far from certainty.
Editorial: there is no body-language decoder here. Patterns over time can be informative (more below), but isolated cues — a look, a laugh, a touch — cannot tell you what someone feels.
THE DATETHOS EDITORIAL FRAMEWORK: Express → Observe → Respond
This framework is our editorial synthesis for organizing the findings above into action — not validated psychology, and not presented as such.
- Express — Communicate your own interest clearly, with warmth, attention, and eventually plain words. This is the part you control.
- Observe — Watch how the other person responds over time: do they reciprocate effort and engage — or withdraw and fade? Treat this as data about the interaction, not a diagnosis of their feelings.
- Respond — Match their engagement. Increase warmth and directness only as interest is reciprocated. Reduce intensity or exit when engagement drops. Stop entirely at any explicit no.
Everything below is an application of these three moves. For the full treatment of the first move, see How to Show Interest and Playfulness.
Express: what showing interest looks like
Research says: question-asking increases liking. Huang et al. (2017) found across live conversations that asking more questions — especially follow-ups — made people better liked, via perceived responsiveness; in speed dating, more follow-ups predicted second-date agreement. (Dependent variables: liking; perceived responsiveness; second-date agreement — a behavioral indicator of liking, not romantic attraction. The speed-dating component was observational.) Editorial: ask genuine follow-up questions — the best-supported conversational behavior here, for liking, not for manufacturing attraction.
Research says: responsiveness matters, conditionally. Birnbaum and Reis (2012) found perceived partner responsiveness was associated with greater sexual desire — primarily in participants low in attachment avoidance; an experiment moved men's attractiveness ratings of a responsive partner, while women's reactions were heterogeneous. (Dependent variables: sexual desire; perceived sexual attractiveness — in a structured, artificial paradigm.) Editorial: be genuinely responsive — listen, validate, show care. Not as a tactic; as how good interaction works.
Research says: reciprocal, escalating self-disclosure builds closeness. Aron et al. (1997): stranger pairs who completed sustained, escalating, reciprocal disclosure tasks reported greater closeness than pairs doing small talk. (Dependent variable: interpersonal closeness — not romantic attraction; participants weren't paired for dating.) Editorial: share gradually and invite the same in return. One-sided oversharing isn't supported here.
Research says: sincere compliments reliably lift the recipient's mood — and givers underestimate this. In Boothby and Bohns (2021), givers underestimated how good recipients would feel and overestimated how bothered they'd be, even in hindsight. (Dependent variable: recipient mood — not liking of the giver, not attraction, not dating outcomes. This was not a dating study.) Editorial: give sincere, specific compliments — on effort, taste, and choices, not just looks — expecting nothing back. The evidence supports kindness here, not attraction engineering.
In live speed-dating, partners rated as funnier were rated more attractive overall — but objectively measured laughter did not predict attractiveness ratings, with no sex differences in live interaction (Wainwright et al., 2023). (Dependent variable: per-date attractiveness ratings — not date outcomes.) Laughing at the other person's jokes was rated effective across sexes and contexts — but those are perceived-effectiveness ratings of hypothetical tactics, not evidence any tactic creates attraction (Kennair et al., 2022). Editorial: enjoy humor together; don't deploy it as a technique.
Editorial: two channels carry much of this — eye contact and touch — each with its own guide, including where the boundaries sit. And for the dynamics of building attraction honestly rather than performing it, see Sexual Tension.
Editorial: flirting over text follows the same framework — express, observe, respond — with thinner information, so ambiguity is higher and explicitness matters more. The medium-specific adjustments live in our texting guide: Flirting Over Text.
Teasing: only if it's respectful — and know it's high-risk
Keltner et al. (1998) found teasers experienced teasing as more positive than targets did — and among couples, flirtatious teasing was higher in those who later broke up. (Dependent variables: liking; positive emotion; breakup vs. persistence.) None of this concerns strangers.
Verbal teasing is a high-risk form of play that emerges later in relationships, not at the start (Baxter, 1992; dependent variable: perceived risk of play forms). Appearance teasing by a potential romantic interest is experienced as hurtful even when recognized as playful (Andrews, 2024; dependent variable: attitudes and reported hurt).
Editorial: teasing is not a flirting technique — the evidence runs against it. If you tease at all, keep it gentle, never about appearance, never punching down, and stop the instant it doesn't land. Full boundaries: Playful Teasing.
Uncertainty is normal — it is not a strategy
Uncertainty about a partner's interest is a documented, normal state — about yourself, the partner, and the relationship (Knobloch and Solomon, 1999; dependent variable: uncertainty subscale scores). Early-stage anxiety is normative too: attachment anxiety runs elevated before relationships are established and predicts approach behaviors (Eastwick and Finkel's [2008] study of attachment in fledgling relationships, JPSP 95; dependent variable: attachment-relevant outcomes). Feeling unsure doesn't mean something is wrong — it means you're at the beginning.
Research says: anxious perceivers misread ambiguity — people high in rejection sensitivity readily perceive intentional rejection in ambiguous behavior (Downey and Feldman, 1996; dependent variable: perceptions of rejection in ambiguous vignettes). Editorial: when you're anxious, your interpretations are least trustworthy. Default to the framework — express, observe, respond — instead of ruminating.
Editorial, with an evidence gap stated plainly: the literature does not support "playing hard to get" or manufacturing uncertainty as a tactic — the studies sometimes cited for it used fictitious profiles and cannot support real-world tactics. Uncertainty is something to tolerate honestly, not weaponize.
Say it plainly — more often than you think
Research says: people use both direct and indirect initiation strategies, with the normative pattern emphasizing direct initiation plus emotional self-disclosure (Clark et al., 1999; dependent variable: strategy use in retrospective self-reports — a catalogue of what people do, not what works). No head-to-head test of direct asking vs. hinting exists in this literature — an evidence gap, stated plainly.
Hinting is how interest gets lost. Indirect signals are the most misreadable channel there is (Abbey, 1982; Hall, 2015) — no validated decoder exists to rescue them. If you want someone to know you're interested, use words: "I'd like to take you out" carries no ambiguity. Directness is also respectful — it gives the other person a clear, answerable question instead of a puzzle. Explicit communication outweighs interpretation, every time.
Analogy, clearly labeled: the sexual-consent literature documents that indirect communication gets dangerously misread and explicit, ongoing communication is safer (Muehlenhard et al., 2016) — a different domain, cited as an analogy for why directness beats hinting, not as dating evidence.
See How to Show Interest for expressing interest directly without pressure. For moving things forward at a pace that respects her responses, see Escalate Respectfully.
Observe: patterns over time, never isolated cues
In bar observations, women displaying high rates of nonverbal solicitation behaviors were approached by men; low-rate signalers were not (Moore, 1985; dependent variable: male approach — not the woman's attraction; observational, confounds uncontrolled). In naturalistic encounters, synchronized laughter tracked mutual reported interest (Grammer, 1990; dependent variable: self-reported interest). These are group-level associations — they don't support "she laughed, so she's interested."
Reciprocity of liking is real — and its dependent variable is liking: people who believed a partner liked them returned greater attraction (Montoya and Insko, 2008). In speed dating, desire was dyadically reciprocal (Eastwick et al., 2007; dependent variable: romantic desire ratings; the proposed "selectivity-signaling" mechanism is contested). Editorial: interest tends to be mutual or not — but "they'll like me because I showed interest" overstates the evidence. Watch whether effort is reciprocated; don't assume you can create it.
Editorial: observe the pattern across interactions, not the moment. Sustained engagement — they initiate, ask about you, make time — is the only "signal" worth acting on, and even then it's evidence about the interaction, not a window into their mind. If your interactions stay friendly but never turn romantic, see Avoiding the Platonic Trap and Friend Zone. For a careful treatment of what limited things can be said about interpreting signals, see Reading Body Language.
When to back off: recognize low engagement and exit
Disengagement usually arrives as withdrawal, not explanation: in a qualitative ghosting study, initiators reported fading out after disinterest, often without explanation (LeFebvre et al., 2019; dependent variable: reported motives — retrospective accounts, not verified causes). Editorial: fading contact is the message — respond to what you can observe, not to guesses about why.
In Baumeister et al.'s (1993) study of unrequited love, rejectors described persistent efforts as intrusive and annoying (dependent variable: coded narratives; retrospective accounts). Editorial: persistence doesn't change minds; it changes how you're perceived — for the worse.
Refusals are systematically softened: when rejecting someone, people report nonpersonal excuses rather than true reasons, to spare feelings (Folkes, 1982; dependent variable: true vs. communicated reasons). Editorial: a softened or indirect refusal is still a refusal. Do not decode it for hidden interest. And an explicit no is absolute: stop, no discussion, no second attempt.
For the complete treatment — calibrating to her responses and exiting gracefully — see Calibration. For handling rejection itself, see our Rejection hub.
Boundaries: the non-negotiables
Editorial — stated as our rules: an explicit no ends it — immediately, with no negotiation and no campaign to change the answer. No touching-escalation systems, no compliance tests, no jealousy plays, no negging, no status games, no sexual pressure, no persistence framed as romantic. None of these appear in the research as effective; several appear as intrusive or harmful. Interest that must be manufactured or extracted isn't interest.
Common flirting mistakes
Editorial synthesis — each grounded in the findings above. For the full catalogue, see Flirting Mistakes:
- Treating a cue as a verdict. One laugh, one look, one touch proves nothing (Hall & Xing, 2015; Barrett et al., 2019). Patterns over time; never isolated cues.
- Assuming friendliness is interest. Warmth is often just warmth (Abbey, 1982).
- Trusting your read when you're most attracted. That's when projection is strongest (Lee et al., 2020).
- Using teasing as a technique. The evidence runs against it with strangers (Keltner et al., 1998; Baxter, 1992).
- Believing compliments create attraction. They reliably lift mood; attraction is untested (Boothby & Bohns, 2021).
- Manufacturing uncertainty. No evidence it works as a tactic; honesty does (evidence gap, stated).
- Persisting past disinterest. Experienced as intrusive (Baumeister et al., 1993); a soft no is a no (Folkes, 1982).
- Hinting instead of saying it. Indirect signals are the most misread channel there is (Hall, 2015).
Where to go next: every flirting guide
Showing interest: How to Show Interest · Playfulness · Flirty Eye Contact · Touch · Playful Teasing
Reading interest: Reading Body Language
Calibration: Calibration · Escalate Respectfully · Avoiding the Platonic Trap · Friend Zone
Dynamics: Sexual Tension
Troubleshooting: Flirting Mistakes