RAADS-R vs AQ vs CAT-Q: Which Adult Autism Test Should You Take?

Deciding to look into whether you might be autistic, as an adult, often starts with a late-night search and ends in confusion. You take one online quiz that says you are “likely autistic,” another that says you are not, and a third built on a test with an intimidating acronym, and you come away less certain than when you began. Part of the problem is that these are genuinely different instruments doing genuinely different jobs, and almost nobody explains the differences. Understanding what each one is for — and what no questionnaire can do — is the difference between a useful first step and a confusing dead end.

One frame to hold throughout: every test on this page is a screening tool, not a diagnostic assessment. Screening tools are self-report questionnaires designed to flag whether autistic traits are present and worth exploring further. A diagnosis is something else entirely — a clinical process involving developmental history, structured observation, and professional judgment, which we cover in depth in our guide to how adult autism is diagnosed. This page is about the questionnaires you will actually meet online, how they differ, and how far to trust them.

§I.Screening questionnaires vs a diagnosis — what these tests really are

The tools covered here all share a basic design: a list of statements you rate for how well they describe you, scored against a threshold derived from research samples. That design makes them fast, free or cheap, and private — genuinely useful for deciding whether to take the question further. It also sets hard limits on what they can tell you. A questionnaire only knows what you report about yourself, in this moment, filtered through your own self-understanding; it cannot watch how you communicate, take a developmental history, or weigh other explanations the way a clinician does. So the right job for any of these tests is narrow and real: help you decide whether a professional conversation is worth pursuing. Held to that job, they are valuable. Asked to deliver a verdict, they will mislead you.

This is also why people end up taking several of these tests in the first place. One screen says one thing, a forum suggests another, and the natural response is to keep testing in search of a certainty a questionnaire was never built to provide. It helps to decide in advance what you are actually looking for — a quick first read, a broad lifespan picture, or a check on how much you mask — and to choose the tool that answers that specific question, rather than collecting scores and hoping they converge on a verdict. They will not, because converging on a verdict is not what they do.

§II.The AQ (Autism-Spectrum Quotient)

The Autism-Spectrum Quotient, published by Simon Baron-Cohen and colleagues in 2001, is the most widely recognized autism screening questionnaire and the one most online quizzes are based on. It asks you to rate agreement with a set of statements spanning five areas associated with the autistic profile — social skill, attention switching, attention to detail, communication, and imagination — and produces a single score, with higher scores indicating more autistic traits. A widely used short form, the AQ-10 (Allison and colleagues, 2012), distills it to ten items as a rapid red-flag screen for whether fuller assessment is warranted.

The AQ’s strengths are its brevity, its enormous research base, and its usefulness as a quick first pass. Its limitations are equally real and important to know. It was developed more than two decades ago against a relatively narrow, stereotype-influenced picture of autism, and it tends to under-detect the people whose autism is least visible — high-masking adults, and women and others whose presentation does not match the older template. A score below the cutoff on the AQ is therefore weak evidence against autism in exactly the population most likely to have been missed for life. It is a reasonable starting screen; it is not the last word, and a low score should not close the question if your experience says otherwise.

It is worth knowing where the AQ came from, because that history explains both its reach and its blind spots. It grew out of a research tradition focused on measurable, observable traits, and it was normed in an era when the prototypical autistic person in the literature was male and outwardly obvious. That lineage makes it an efficient, heavily studied screen and a poor net for the inward, camouflaged presentations that dominate late-diagnosed adults. If you score low on the AQ but the rest of your experience — the exhaustion of socializing, the lifelong sense of running different software than everyone around you — points the other way, the sensible reading is that the AQ may simply be the wrong instrument for your presentation, not that the question is closed.

§III.The RAADS-R

The Ritvo Autism Asperger Diagnostic Scale–Revised, published by Riva Ritvo and colleagues in 2011, is a longer questionnaire — eighty items — that was designed as a clinical aid to help identify autism in adults who may have been missed in childhood. It covers four areas (social relatedness, circumscribed interests, language, and sensory-motor experience) and, unlike a snapshot screen, asks you to consider traits across your lifespan, including childhood, which is a genuine strength for adult self-investigation. Many adults find it the most thorough and recognizing of the questionnaires — it asks about experiences the briefer screens skip entirely.

Its central limitation is one you should weigh carefully: the RAADS-R was validated in specialist clinical settings, administered alongside professional judgment, and its performance is very different when taken cold, online, by someone who already suspects they are autistic. In that self-administered, general-population context it produces a high rate of false positives — scores above the cutoff in people who are not, in fact, autistic. This is not a flaw in the test so much as a mismatch between how it was built to be used and how it is used in practice. A high RAADS-R score taken on your own is a meaningful prompt to seek assessment; it is not, despite the official-sounding name, a diagnosis or anything close to one.

What many adults appreciate about the RAADS-R is precisely its length and its reach back into childhood. Because identifying autism in adults depends heavily on establishing that traits were present early in life, a questionnaire that explicitly asks you to consider how you were as a child — not only how you are now — captures something the snapshot screens cannot. The trade-off for that thoroughness is time, and the false-positive issue already noted: the very breadth that makes it feel recognizing also makes it easy to clear the threshold on the strength of traits that, taken in isolation, are not specific to autism. Thorough and recognizing is not the same thing as conclusive.

The 97% in the brochure was measured in a best case you’re probably not in.

§IV.The CAT-Q — the one that measures masking

The Camouflaging Autistic Traits Questionnaire, developed by Laura Hull and colleagues in 2019, is fundamentally different from the other two, and understanding why is the key to this whole comparison. The AQ and RAADS-R try to detect autistic traits. The CAT-Q does not measure autism at all — it measures camouflaging, the conscious and unconscious strategies people use to hide or compensate for autistic traits in order to appear non-autistic. Its twenty-five items map onto three kinds of camouflaging: compensation (actively working around social difficulties), masking (hiding autistic characteristics), and assimilation (forcing yourself to fit in).

Why does this matter so much? Because masking is the single biggest reason the trait-based screens fail in the people most often missed. If you have spent your life learning to perform neurotypically — scripting conversations, copying expressions, suppressing stims — a trait questionnaire asking what you are like on the surface may score you as barely autistic, precisely because you have become so good at hiding it. The CAT-Q catches what the others miss: a high camouflaging score alongside a borderline AQ is one of the most telling patterns in adult self-investigation, especially for women and high-maskers. The trade-off is that the CAT-Q only tells you about masking; it cannot, on its own, tell you whether you are autistic. It is best read together with a trait measure, not instead of one. (We go deep on this in our guide to autistic masking.)

§V.RAADS-R vs AQ vs CAT-Q: how to choose

There is no single “best” test, because they answer different questions. The useful move is to match the tool to what you are trying to learn:

If you want one practical recommendation: no single questionnaire taken alone gives a trustworthy picture, and the most informative approach is to read a trait measure and a masking measure together — because the gap between “how autistic I look” and “how hard I am working to look non-autistic” is often where the real answer lives. That combined logic — traits plus masking plus emotional context, built for adults rather than retrofitted from older tools — is exactly what our own inventory was designed around.

§VI.What your score actually means

Every one of these tests reports a number against a threshold, and almost everyone over-reads that number. Here is how to read it honestly. A score above the cutoff means your self-reported traits (or masking) are consistent with the autistic range and that further exploration is reasonable — it does not mean you have been diagnosed, and given the false-positive issues above, it is a flag, not a finding. A score below the cutoff means your reported traits did not reach the threshold on that instrument — which, in a high-masker or someone the test was not designed to catch, can be a false negative rather than reassurance. The specific numbers (a particular RAADS-R total, a particular AQ score) are best understood as positions on a continuum, not as pass/fail lines. Two people just above and just below a cutoff are far more similar than the line between them suggests.

A concrete picture helps. Imagine two people who take the same screen and land a single point apart — one just above the cutoff, one just below. The instrument sorts them into “screen positive” and “screen negative,” and it is tempting to treat that as a real boundary between autistic and not. But one point on a self-report scale is well within the noise of how a person happens to interpret a few ambiguously worded items on a given day. The two of them almost certainly have more in common with each other than either does with someone at the far end of the range. The cutoff is a useful administrative line for deciding whom to look at more closely; it is not a fact about the people on either side of it.

The single most important thing to do with any score is to hold it lightly and in context. A questionnaire result is one data point about how you see yourself today. It earns its value when it prompts a good next step — reflection, more information, or a professional conversation — and it does harm when it is treated as a definitive label in either direction.

A screener answers one question: “should I look into this?” It never answers “do I have it?”

§VII.Are online autism tests accurate?

This is the question almost everyone is really asking, and the honest answer has two parts. As screening tools used as intended, the validated questionnaires are genuinely useful — they reliably flag traits worth exploring, and a thoughtful self-test can be the thing that finally puts language to a lifetime of feeling different. As diagnostic tools, no online test is accurate, because diagnosis is not something a self-report questionnaire can do, no matter how well-designed. Accuracy, in other words, depends entirely on what you ask the test to do.

Two cautions sharpen this. First, the popular online versions vary enormously in quality; many are unvalidated knock-offs of the real instruments, and some are outright unreliable, so the source matters as much as the test. Second, even the validated questionnaires were generally normed in clinical or research samples and behave differently when self-administered by motivated searchers — which is why false positives (and, for high-maskers, false negatives) are common. A good self-test is an accurate screen and an inaccurate oracle. Use it for the former.

The practical upshot is to be as careful about where you take a test as about which one you take. A validated questionnaire reproduced faithfully, scored correctly, and interpreted as a screen is a legitimate tool; the same instrument mangled into a clickbait quiz with invented scoring and a confident “you are autistic” verdict is worse than useless, because it manufactures false certainty in both directions. When a result matters to you, prefer sources that are transparent about what the test measures, say where it came from, and tell you plainly that it cannot diagnose — and treat anything promising a definitive answer as a reason for more skepticism, not less.

§VIII.From a high score to an answer

If a screen points toward autism and that resonates, the next step is not another quiz — it is deciding whether to pursue a formal assessment, and understanding what that involves. A professional evaluation can do what no questionnaire can: take a developmental history, observe communication directly, weigh other explanations, and integrate it all into a judgment. Some adults want that formal answer for access to support, accommodations, or simple certainty; others find that self-identification, grounded in honest self-assessment and reading, gives them what they need. Both are legitimate paths, and we walk through the trade-offs, the process, and how to find a qualified assessor in our guide to how adult autism is diagnosed. The questionnaires on this page are the on-ramp to that decision, not a substitute for it.

§IX.Levels of autism (DSM-5): what the tests don’t tell you

People often expect a test to tell them “what level” of autism they have. It cannot, and the levels are worth understanding on their own terms. The current diagnostic manual (the DSM-5-TR) describes autism as a single spectrum and assigns a support level from one to three based on how much support a person needs in two areas, social communication and restricted/repetitive behaviors. Level 1 indicates the lowest support needs (often the adults who reach midlife undiagnosed); Level 2 indicates substantial support needs; Level 3 indicates very substantial support needs.

Two things to keep in mind. First, the levels describe support needs in context, not severity of the condition, intelligence, or how much a person is struggling internally — a “Level 1” adult can be in significant distress that simply is not visible. Second, levels are assigned by a clinician during assessment, can differ across the two domains, and can shift with circumstances and support; no self-report questionnaire produces or replaces them. The AQ, RAADS-R, and CAT-Q measure traits and masking, not support levels — so if a quiz claims to tell you your “autism level,” that alone is a sign it is not a serious instrument.

§X.“High-functioning autism” and Asperger’s

You will run into both terms constantly, so it helps to know where they stand. Asperger’s syndrome was a separate diagnosis in older manuals, used for autistic people without language delay or intellectual disability. It was folded into the single diagnosis of autism spectrum disorder in 2013, so it is no longer a separate clinical diagnosis — though many people diagnosed under the old system, and many cultures and communities, still use the word, and that is a personal choice worth respecting.

“High-functioning autism” is an informal label, not a clinical category, generally used to describe autistic people who can live and work relatively independently and whose traits are less outwardly obvious. Clinicians and many autistic adults have moved away from it because it conflates two different things — how autism appears to others versus how much a person is actually struggling — and because “high-functioning” on the outside is very often the product of exhausting, costly masking on the inside. The traits these terms point to are exactly the ones the legacy screens under-detect, which is part of why so many “high-functioning” adults go unrecognized for decades. If that pattern sounds familiar, our guides to the signs of autism in adults and autism in women go deeper.

§XI.Where the LifeByLogic Self-Inventory fits

We built the Adult Autism Self-Inventory because of the gaps this whole guide has been describing. The most widely used screens are decades old and under-detect the people most likely to have been missed; the masking measure that catches them is rarely combined with anything else; and almost none of these tools were designed from the ground up for adults investigating themselves. Our inventory is an original, source-cited instrument that covers social processing, sensory sensitivity, routine and change, and masking together — the combined picture the legacy tests fragment — and it is written for the adult, late-recognition experience rather than retrofitted from older, narrower templates.

Two honest boundaries. First, it is just as non-diagnostic as every other tool on this page — it cannot diagnose you, and it is not trying to; only a qualified professional can do that. Second, because it is LBL-original rather than a reproduction of a copyrighted clinical scale, it is a screen and a structured reflection, designed to give you a clearer, more complete starting picture and a sensible next step — not to stand in for an assessment. Used for what it is, it is the single most useful starting point we know how to offer.

§XII.Common questions

Which autism test is most accurate for adults? No single questionnaire is definitively “most accurate,” because they measure different things and all are screens rather than diagnoses. For adults — especially women and high-maskers — reading a trait measure together with the CAT-Q (a masking measure) gives a fuller picture than any one test alone, and a formal assessment is what actually settles the question.

Is the RAADS-R accurate, or does it give false positives? The RAADS-R is a thorough, clinically derived questionnaire, but when taken alone online by people who already suspect they are autistic, it produces a high rate of false positives. A high score is a strong prompt to seek assessment, not a diagnosis.

What is a high CAT-Q score, and does it mean I’m autistic? A high CAT-Q score means you engage in a lot of camouflaging or masking. On its own it does not establish autism — it measures masking, not autism — but a high masking score alongside borderline trait scores is a common and telling pattern in adults who were missed, and a good reason to look further.

Can an online autism test diagnose me? No. No online questionnaire can diagnose autism. Diagnosis requires a clinical process — developmental history, direct observation, and professional judgment. Online tests are screening tools that tell you whether seeking that assessment is worthwhile.

Can these tests tell me what “level” of autism I have? No. DSM-5 support levels are assigned by a clinician during assessment and describe support needs, not severity. Self-report questionnaires measure traits or masking, not levels — any quiz claiming to output your “level” is not a serious instrument.

Is “high-functioning autism” or “Asperger’s” a real diagnosis? Neither is a current clinical diagnosis. Asperger’s was merged into autism spectrum disorder in 2013, and “high-functioning” is an informal label many have moved away from because it hides how much a person may be struggling beneath a capable surface.

§XIII.Where to start

If you are at the beginning of this, the most useful first step is a single, honest screen that gives you a complete starting picture rather than a fragment — one that accounts for masking, not just surface traits — and then a clear sense of whether to pursue a formal assessment. That is what our self-inventory is for: it stores nothing, diagnoses nothing, takes about ten minutes, and is designed to leave you understanding your own pattern and your options, not just holding a number.

Whatever you take — ours, the AQ, the RAADS-R, the CAT-Q, or several — hold the result as a starting point, read your trait and masking picture together, and remember that the question a questionnaire opens is best answered by a person, not a score.