Where did 'words are only 7%' come from?
You have probably heard it: 'In communication, the words are 7%, tone of voice is 38%, and facial expression and gesture are 55%.' This '7-38-55 rule' is a staple of presentation courses, interview tips, and self-help books, always cited with the name of UCLA psychologist Albert Mehrabian. Yet this famous set of numbers is quite different from what the original research actually said.
The numbers themselves are real. They appear in Mehrabian's 1971 book Silent Messages, built on two experiments he published in 1967. The trouble is that the figures got stretched into claims like '93% of all conversation is nonverbal,' while the narrow conditions the experiments actually studied were quietly cut away.
This is a piece of impression psychology that tries to clear up that misreading. It is not here to put anyone down or to argue that words do not matter — it simply walks calmly through what these often-quoted numbers really meant.
What the experiments actually measured — like and dislike
Mehrabian's two experiments looked at one thing: when words and nonverbal cues conflict, which does the listener believe? One study (Mehrabian & Wiener, 1967) played the same word recorded in different tones of voice; the other (Mehrabian & Ferris, 1967) paired spoken words with photos of different facial expressions. A neutral word like 'maybe,' for instance, would be delivered in a warm tone and then a cold one.
What was measured was not how much the content of words counts, but how much people leaned on each signal when inferring whether someone liked or disliked them. Mehrabian summarized the result as 'total liking = 7% verbal + 38% vocal + 55% facial.' The key point is that this is a weighting for when the cues to liking contradict one another.
The narrow conditions the formula applies to
So 7-38-55 comes with two conditions. First, it applies only when the subject is feelings and attitudes (like versus dislike). It does not cover informational speech, like giving directions or explaining a fact. Second, it is about cases where words and nonverbal cues conflict. When they agree, there is no need to weigh which to trust.
Seeing the study's limits too
The study's limits are clear as well. The participants were few and all women, and it was a lab setup using recordings and photos rather than real conversation. And although it is called 'nonverbal,' it really covered only facial expression, not whole-body movement. So carrying these numbers straight into every talk, negotiation, or daily chat does not hold up.
The warning Mehrabian himself gave
Tellingly, the person who pushed back hardest against the misreading was Mehrabian himself. He made clear that the equation came from situations communicating feelings and attitudes, and that unless a communicator is discussing their feelings or attitudes, these equations are not applicable. In interviews he has said he winces whenever he hears the 'only 7% of the message is the words' misquote.
In other words, the slogan '93% of conversation is nonverbal' is an over-generalization the original author never intended. It is a classic case of good research being compressed into a short, striking number while the narrow context it lived in disappears.
| Channel | Meaning in the study | Weight (inferring liking, conflicting cues) | Common myth |
|---|---|---|---|
| Words (verbal) | Cue to liking carried by the word | About 7% | 'The content of words is only 7%' — not true |
| Voice (tone) | Cue to liking carried by tone | About 38% | Applies to all conversation — it does not |
| Face (expression) | Cue to liking carried by expression | About 55% | Measured all body language — only the face |
| Premise | When feelings and attitudes conflict | — | Works for informational talk too — it does not |
So how is an impression formed? — a balanced view
That does not mean expression and tone do not matter. What Mehrabian's research really shows is that when words clash with face and tone, people tend to trust the nonverbal side. The real lesson is congruence. Say something warm with a cold expression and the message wobbles; when all three point the same way, the impression comes through clearly.
Applied to first impressions: what you say matters, and so does keeping your words, expression, and voice from contradicting each other. Rather than memorizing that any one of them is 93%, it is genuinely more useful to remember that gathering the three signals in the same direction makes an impression sharper.
A small note for photos and video
A photo has no voice and no flowing speech. So in a profile photo, expression and posture carry a large share of the 'feeling.' You cannot import Mehrabian's numbers directly, but the principle — keep the mood you want to convey from clashing with your expression — works the same in a photo. If you want an easy-going impression, relax the shoulders and the area around the eyes together.
Still, all of this is about impression. A good-looking expression in a photo does not tell you a person's character or ability, and it is not used to claim that. Please just remember that impression is something to enjoy shaping, not a yardstick for judging anyone.
Frequently asked questions
So the content of words does not matter?
Not so. 'Words are only 7%' is the classic misreading. Mehrabian's numbers only show that when feelings and attitudes clash with the words, people lean on the nonverbal cues — and in informational conversation, the content of the words is of course central.
Can I apply 7-38-55 directly to talks or interviews?
Not directly. The original studies covered the narrow case of feelings — like and dislike — conflicting with the words. In talks and interviews, content, structure, expression, and voice all matter, and above all, whether the three point the same way shapes the impression.
Why did these numbers get so famous?
They are short, easy to remember, and carry the surprise of 'nonverbal matters that much?!' As they were compressed into one striking line, the original narrow conditions fell away, and a slogan that differs from the real meaning spread widely.
A photo has no voice, so how is this relevant?
It does not apply directly, but the principle carries over. In a photo, expression and posture largely shape the 'feeling,' so keeping the mood you want and your expression from clashing makes the impression clearer. Of course, a good impression does not tell anyone's character.
Article info & references
Published July 2, 2026 · Last updated July 2, 2026
- Mehrabian, A. (1971). Silent Messages. Wadsworth. (the source of 7-38-55)
- Mehrabian & Wiener (1967), 'Decoding of Inconsistent Communications'; Mehrabian & Ferris (1967), 'Inference of Attitudes from Nonverbal Communication in Two Channels' (the two original studies)
- General references on Albert Mehrabian and his own caution about the equation's scope (en.wikipedia.org, 'Albert Mehrabian')
- General social-psychology concepts on impression formation, such as the primacy effect and message congruence