A widely repeated statistic — that communication is something like 93% nonverbal, or that body language reveals what someone “really” means regardless of their words — gets used to justify treating crossed arms, eye contact, or posture as a more trustworthy signal than what someone actually says.

What the evidence actually supports

The specific statistic behind this claim has been badly overextended from its original, narrow context. The research it traces back to concerned emotional expression in a specific, limited kind of message — not general communication, and not a claim that body language overrides or is more truthful than words in everyday interaction. The popularized “93% nonverbal” version is a widely repeated misreading of a much narrower finding (established evidence — this specific misapplication of the original research is well documented and widely corrected within psychology).

Some body language does carry real, if limited, information. Genuine discomfort, disengagement, or openness can show up in posture, gesture, and expression, and reading these cues is a real, useful skill — just not one that operates with anything close to the precision popular claims suggest (established evidence — nonverbal cues carry real information; established evidence that the specificity and reliability of that information is far more limited and context-dependent than commonly claimed).

A single nonverbal cue, read in isolation, is weak and easily misread. Crossed arms can mean discomfort, or cold, or simple habit. A lack of eye contact can mean disinterest, or cultural norms, or attentiveness expressed differently. The same cue routinely means different things in different people and different contexts (established evidence).

Cues become more informative in combination and in context, never in isolation. Multiple consistent signals, read against someone’s own baseline behavior, carry real evidentiary weight — the same principle covered throughout this library for any kind of evidence-weighing, applied here specifically to nonverbal cues (established evidence for the general principle; DEZAIR synthesis for its specific application to body language).

The honest uncertainty

There’s no reliable, universal decoder for body language — no fixed rulebook where a specific gesture always means a specific thing across different people, cultures, and contexts. Claims that promise this level of precision are overstating what the underlying research actually supports.

What this actually means in practice

Treat body language the same way this library treats every other kind of evidence: real, worth attending to, and never sufficient on its own. A single cue is a prompt to look for more information, not a conclusion. Multiple consistent cues, read against a person’s own baseline rather than a generic rulebook, carry genuine weight. And when a nonverbal cue conflicts with something someone has said directly, that conflict is itself worth a direct check-in — not an automatic reason to trust the nonverbal read over the spoken one, or vice versa.

Claim strength, at a glance

  • The “93% nonverbal” statistic as a general claim about communication: not supported — a well-documented overextension of narrower original research
  • Nonverbal cues carry some real information: established evidence
  • A single cue, read in isolation, is weak and easily misread: established evidence
  • Multiple cues in context and combination carry more weight: established evidence for the general principle

Where to go next

For the broader principle this applies — one signal is noise, patterns are information — see One Signal Is Noise. Patterns Are Information. For what this means during an actual intimate encounter specifically, see Why Silence During Intimacy Is Hard to Interpret.