How do I know if someone I'm seeing is lying?

Mostly, you can’t tell by watching them, and the research on that is humbling: people spotting lies do only a little better than a coin flip.1 What actually works is slower and less dramatic. You notice whether what they tell you keeps matching what happens, and you ask about the part that doesn’t.

Why can’t you just read it on their face?

Picture Camila, 24. Her last boyfriend spent most of a year “working late,” and he wasn’t. Now she’s three weeks into seeing Rafi, who is funny and texts back and seems kind. On Thursday he says he can’t do Friday because of “a family thing.” When she asks what kind, he says “just my mom stuff” and changes the subject. Camila spends the drive home replaying it. Did he look away? Did he answer too fast? Was the subject change on purpose?

Here’s the problem with that replay. In a 2006 meta-analysis pooling 206 documents and 24,483 people judging whether someone was lying, Charles Bond and Bella DePaulo found average accuracy of 54%. People correctly caught 47% of lies and correctly believed 61% of truths.1 That’s barely better than guessing, and those judges weren’t more hopeful or more tired than you are on a Thursday night.

It isn’t that people miss obvious signs. The signs mostly aren’t there. A 2003 review led by DePaulo combined 1,338 estimates of 158 possible cues to lying. Liars were a bit less forthcoming, told somewhat less compelling stories and seemed more tense, but many behaviors showed no link or only a weak link to lying at all.2 Even when researchers do find traces, they’re subtle. A study of 78 online daters found that more deceptive profiles used fewer self-references, more negations and fewer words overall in their written bios.3 That pattern showed up when researchers analyzed the language across the whole group, not in any one bio. Looking away, fidgeting and answering fast are not a lie detector. Neither is your gut reading a text bubble.

So what does give a lie away?

Mostly time and facts, not tells. A story either holds together as more of it becomes visible, or it starts to fray. Things worth paying attention to:

• Details that collide. He said family Friday, then his friend’s story shows him at a bar. That’s a real mismatch, and it deserves a real question. It still isn’t a verdict, because there may be an ordinary explanation (the family thing ended at nine, say).
• The same story, told twice. What someone says about their job, their ex or their weekend should sound roughly the same on date five as on date two.
• What happens when you ask. A plain question usually gets a plain answer. Someone who gets angry or slippery every time you ask a normal thing is showing you a pattern worth weighing.

Notice that none of these require you to read anyone’s mind. They’re about what you can actually see.

Why does everything feel suspicious right now?

Because something real happened to you. When you were lied to, the alarm that should have gone off didn’t, so now it goes off early, at everything. That’s understandable. It’s also exhausting, and it can make a new person pay for the old one.

It helps to pull Camila’s Thursday apart into four pieces:

• What happened: Rafi said “family thing” and didn’t give details.
• The reading: he’s hiding something.
• The fear: I’m going to find out in eight months, like last time.
• The next step: see whether Friday’s story stays the same, and ask if it doesn’t.

Only the first piece is something Rafi did. The reading came from her ex. The fear is real and deserves kindness, but it belongs to her history. The next step is the only part she can act on, and it’s small.

What about the small stuff everyone does?

Some fibbing is almost background noise. In a field study of 183 people who text, researchers at Cornell described the plausible, sometimes untrue stories people tell about where they are and what they’re doing, including “butler lies” used to start, end or dodge a conversation.4 “Just saw this!” and “omw” when you haven’t left are the classics. They’re not great, but they’re social smoothing, not deception about the relationship.

Lies inside romantic relationships also come from more than one place. In a study of deception between romantic partners, communication researchers Guthrie and Kunkel found people deceive partners to keep the relationship going, to save face, to manage tensions between them and sometimes to gain control.5 The motive matters. “I loved the playlist” and a story that hides who they’re seeing are not the same act.

The apps add their own layer. In a 2019 Pew Research Center survey, roughly seven in ten online daters said it’s very common for people on these platforms to lie to seem more desirable.6 A flattering photo or a generous height is annoying. A lie about being single, about other people they’re seeing or about something that affects your health or money is a different category. Keep the categories separate, or you’ll wear yourself out treating every rounded-up detail like a betrayal.

How do you ask without it turning into an interrogation?

Ask one thing, about one specific detail, out of curiosity rather than as a closing argument. “I saw you out Friday, was the family thing early?” lands very differently from “So were you even at your mom’s?”

Then pay attention to two things: the answer, and how they handle being asked. Someone who explains easily, or says “oh, that came out wrong,” is giving you information. Someone who turns your question into proof that you’re crazy is also giving you information.

What should you do this week?

• Write down what you saw, not what it means. Two columns, if it helps: “happened” and “my reading.”
• Ask one question you’ve been sitting on. One. In person or on a call if you can.
• Stop the friend panel. Screenshots sent around for verdicts mostly make the story louder. Your friends can’t read tone in a text any better than you can.
• Give it a few weeks of watching what matches. Trust builds from stories that stay the same and plans that actually happen.
• Notice when it’s really about the last one. If every vague answer sends you back to the same memory, that memory may need more attention than this person does.

If the suspicion follows you into every new connection, keeps you up at night or makes you avoid dating for months, a licensed mental-health professional is the right kind of help. Being lied to by someone close can leave a long shadow, and you don’t have to sort it out alone.

What doesn’t the research show?

• Lie-detection studies mostly use strangers in labs, often judging short clips. Real relationships give you more context, which is exactly why facts over time beat tells.
• The cue review averages many studies, so it can’t say whether one particular person shows a particular tell.
• The dating-profile study looked for language patterns across many profiles, which is not something you can do reliably by reading one person’s texts.
• The romantic-deception research describes why people say they lie, not how often the person you’re seeing does.
• The texting study describes everyday availability fibs, not lies about relationships or other partners.
• The Pew figure is a perception of how common lying is on the apps, not a measured rate of lying.