The free MBTI test

Everything above runs in your browser. Nothing is emailed to you, nothing is held back, and there is no “unlock your full report” step at the end — the full report is the type page your result links to. What follows is what the questionnaire is actually doing, and where it stops being trustworthy.

What this MBTI test measures

This questionnaire is independent: it is not affiliated with, endorsed by, or derived from The Myers-Briggs Company, 16Personalities, or any official MBTI^® instrument. The four dichotomies are a publicly described framework; these 48 items are our own wording.

Four preference pairs, and nothing beyond them. Each of the 48 statements loads onto exactly one pair, and the four winners get concatenated into your code.

• E / I — where your attention goes. Whether time around other people tops you up or draws you down, and whether you think out loud or turn up with the thought already finished.
• S / N — which information you trust first. The concrete, checkable version of a situation, or the pattern you suspect it belongs to.
• T / F — what you check a decision against. Whether the reasoning holds together on its own terms, or what the decision does to the people standing inside it.
• J / P — how you relate to closure. Whether settling something is a relief that frees you up, or a door closing on options you would rather keep open.

That is the whole instrument — and it is worth being precise about where cognitive functions fit, because this is the most misunderstood part of the model. Jung’s functions are where the theory comes from, but no MBTI questionnaire has ever measured them: the official instrument scores these same four dichotomies, and the familiar Ni-Te-Fi-Se stack is derived from the four letters by a fixed rule rather than tested for. We show that derived stack with your result, because the derivation is standard and deterministic. What it is not is independent evidence — it inherits every error in your letters, so a borderline J/P makes for a borderline stack. Untangling which functions you actually run takes more than twelve agree-or-disagree statements per axis; for that, work through the cognitive functions guide deliberately. If the four letters themselves are new to you, the MBTI guide covers where they came from and what they were originally meant to describe.

How this test is scored

Openly, because the mechanics are the best argument for holding your result loosely.

There are 48 statements, twelve per dichotomy, shown six to a page. Within each set of twelve, six are written so that agreeing pushes toward one pole and six so that agreeing pushes toward the other. That balance is deliberate: plenty of people have a mild habit of agreeing with whatever sentence is in front of them, and mirrored wording cancels most of that out instead of letting it accumulate into a letter.

Each answer is worth between −2 and +2. The answers on a dichotomy are multiplied by their direction and added up, producing a raw score somewhere between −24 and +24, which is then rescaled to a 0–100 lean. Whichever side of the midpoint you land on takes the letter. Anything you skip counts as a neutral zero, so a half-finished test still returns a result — just a weaker one.

Two things that percentage is not. It is not a percentile: there is no norming sample behind this test, so “68% toward Intuition” means you leaned that way across these particular items, not that you out-scored 68% of anyone. And it is not a confidence rating. That is why the result also names your runner-up type — who you would have been if the narrowest of the four calls had gone the other way — and flags any call that landed near the middle rather than quietly rounding it into a letter.

All 16 MBTI types

Your result drops you straight into your own profile, but the library is open regardless of what you scored. Each page covers how that type argues, writes, works and mistypes.

How accurate is a self-report MBTI test?

Accurate enough to be a useful hypothesis, not accurate enough to be a verdict — and that is true of every dichotomy-based questionnaire, this one included. Three well-documented weaknesses, none of which a free test escapes:

• Retest drift. People who sit the same style of questionnaire twice, a few weeks apart, quite often come back with at least one letter changed. That is not a defect in one particular quiz; it falls out of the next problem.
• A binary drawn across a continuum. Scores on these four dimensions spread out like one hill, not two separate camps — most people sit somewhere in the middle rather than out at a pole. A four-letter code has to plant a fence somewhere on that hill, so anyone standing near the fence gets sorted by a couple of points. If your result flagged a dichotomy as close, that is exactly what happened, and the runner-up type is a live option, not a consolation prize.
• You are the measuring device. Every answer was a judgement you made about yourself, which imports whatever you currently believe about yourself, whatever you happened to be in the middle of when you sat down, and — quietly — whoever you would prefer to be. A form has no way to tell an accurate self-report from a flattering one.

This is also why the honest version of “is my type right?” is not a number. It is whether the profile describes someone you recognise from your own worst week, or someone you would enjoy being introduced to. Read your type page with that question in hand.

This test vs. reading what you already wrote

There is a version of this that skips the questionnaire, and the gap between the two is sharpest exactly where the questionnaire is thinnest: function order. The stack you get above is read off your letters. A stack read off your behaviour is a different claim entirely.

Cognitive functions show up in behaviour rather than in self-assessment — in whether you build the frame before the details or after, in what you reach for when you are cornered, in which register you drop into when you are tired and typing fast. All of that is legible in writing you produced for real reasons, and structurally invisible to a form asking you to rate your own preference for structure. TypeMyVibe reads the writing and infers a function order from it, which is a thing a Likert scale cannot do at any length.

The second difference is cross-checking. Four letters on their own have nothing to argue with. When MBTI, Enneagram, Big Five, Attitudinal Psyche and Alignment are all read off the same text, they either converge — which is informative — or they disagree in one specific place, which is more informative still. The walk-through in the AI MBTI test shows what that read looks like end to end, including which lines produced which call.

Try the other four free tests

Same arrangement: written from scratch, scored in your browser, no signup. Each one sees something the four letters cannot.

• Enneagram test — MBTI describes how you process; the Enneagram describes what you are trying to avoid. Two people with identical letters can be running completely different fears.
• Big Five (SLOAN) test — the trait model academic psychology actually uses, condensed into a 32-way letter code. If it is MBTI’s scientific standing that bothers you, start here instead.
• Attitudinal Psyche test — ranks Volition, Logic, Expression and Physics into a stack of four. It is the closest thing to MBTI’s function-order idea with a scoring method you can follow line by line.
• Moral alignment test — order against improvisation, self-interest against concern for others. Not a personality model, but it explains a category of friction the other four keep missing.

All five sit together on the free tests page if you would rather pick from the shelf.