RAADS-R Test: Online Test for Autism – Take Test and Understand Scoring
A clear guide to the RAADS-R test. What the 80 questions look like, how the 0 to 240 score is worked out, what every score band means, and what happened when researchers tested it outside the lab.
Quick answers
| Full name | Ritvo Autism Asperger Diagnostic Scale, Revised |
| Questions | 80 |
| Time | 15 to 20 minutes |
| Score range | 0 to 240 |
| Standard cutoff | 65 |
| Subscales | Four |
| Published | 2011, by Ritvo and colleagues |
| Designed for | Adults aged 18 and over |
| Cost | Free for personal, clinical and research use |
| Can it diagnose? | No |
What the RAADS-R test is
The RAADS-R test is an 80-item questionnaire. It measures autistic traits in adults. It was published in 2011 by Riva Ariella Ritvo, Edward Ritvo and colleagues.
It was built for a specific group. Most autism instruments came from research on children. That made them awkward for a forty-year-old who had spent decades quietly adapting. The RAADS-R test was written for that person.
One feature sets it apart. It asks about your whole life, not just now. Autism is developmental by definition, present from early childhood even when nobody spotted it. A questionnaire that only asks about this month can miss someone completely.
How the questions work
You read a statement. You pick one of four options. Each option describes when that statement has been true for you.
Those numbers are the points. Most items score this way. A smaller number are reversed, so the scoring runs the other direction.
The reversal matters. Without it, anyone who agreed with everything would score 240. Reversed items break that pattern.
Sample questions
The actual RAADS-R test items are copyrighted by their authors and licensed for non-commercial use, so they are not reproduced here. The five statements below are original. They are written to show you the style, the subject matter and the response format you will meet.
- Conversations make more sense to me afterwards than during.
Social relatedness. The largest subscale, covering conversation, friendship and reading intent. - I notice a hum or a flicker in a room long before anyone else mentions it.
Sensory-motor. Covers sound, light, touch, texture and repetitive movement. - When someone says something is “a piece of cake”, I picture the cake first.
Language. The smallest subscale, covering literal interpretation and figures of speech. - I can talk about one subject for hours without noticing the other person has stopped listening.
Circumscribed interests. Depth of interest, detail focus and routine. - I have always needed to know the plan before I can relax into an event.
Social relatedness again, this time the predictability and routine side of it.
Notice the phrasing of the fifth one. “I have always” points at the lifetime framing. Roughly half the real items are worded to pull your attention back to childhood.
How scoring works
Add up the points from all 80 items. The result sits somewhere between 0 and 240. That is the whole method.
The published threshold is 65. Score above it and the original paper described your result as consistent with autism. Score below and it described autism as unlikely.
That single number hides a lot. A score of 66 and a score of 200 both sit above the line, yet they describe very different people. This is the main criticism of how the RAADS-R test is usually reported.
What the numbers looked like in the original study
| Group | Approximate mean score |
|---|---|
| Autistic adults | Around 134 |
| Comparison group | Around 26 |
| Threshold | 65 |
Those averages are more useful than the cutoff. A score of 70 clears the threshold but sits far below the autistic group average. A score of 140 sits above it.
Full score lookup table
Use this to place a score in context. The bands are descriptive. None of them is a diagnosis.
| Score | Band | What it generally means |
|---|---|---|
| 0 to 25 | Very low | Around the average for non-autistic adults. Few traits reported. |
| 26 to 49 | Low | Some traits, well below the threshold. Common in the general population. |
| 50 to 64 | Below threshold | Elevated but under the cutoff. Worth looking at the subscales. |
| 65 to 79 | At threshold | Just clears the published cutoff. The weakest evidence of any band above the line. |
| 80 to 104 | Moderate | Clearly above threshold. Still well below the autistic group average. |
| 105 to 129 | Substantial | Approaching the average score of diagnosed adults in the original study. |
| 130 to 159 | High | Around and above the autistic group average of roughly 134. |
| 160 to 189 | Very high | Well above the autistic group average. Traits reported across all four areas. |
| 190 to 240 | Extremely high | Near the ceiling. Uncommon even among diagnosed adults. |
The simplified four-band version
Later work proposed a coarser set of bands, because the single cutoff was failing in clinical settings. You will see these quoted too.
| Score | Interpretation |
|---|---|
| Below 65 | Low level of autistic traits |
| 65 to 105 | Some autistic traits |
| 106 to 139 | Traits consistent with autism |
| 140 and above | Pronounced autistic traits |
These revised bands report sensitivity and specificity of roughly 81 per cent. That is a long way from the original claim, and considerably more honest.
Subscale scores explained
The 80 items split across four areas. They are not equal in size. Social relatedness carries almost half the test.
| Subscale | Items | Max score | What it covers |
|---|---|---|---|
| Social relatedness | 39 | 117 | Conversation, friendship, reading intent, social reciprocity, routine |
| Sensory-motor | 20 | 60 | Sound, light, touch, texture, repetitive movement |
| Circumscribed interests | 14 | 42 | Depth of interest, detail focus, collecting and ordering |
| Language | 7 | 21 | Literal interpretation, figures of speech, pragmatic language use |
| Total | 80 | 240 |
Compare subscales as percentages, not raw numbers. A language score of 15 looks small next to a social score of 60. As proportions they are 71 per cent and 51 per cent, which tells a different story.
How accurate it is
This is where most pages stop short. They quote the 2011 figures and move on. The full picture is more interesting.
What the original study reported
Ritvo and colleagues tested 201 autistic adults against 578 comparison participants. At a cutoff of 65 they found 97 per cent sensitivity and 100 per cent specificity. Test-retest reliability was 0.987. Those are remarkable numbers.
What happened when others tested it
The picture changed sharply. A 2021 UK study gave the RAADS-R test to 50 adults waiting for assessment at a specialist service. Sensitivity was 100 per cent. Specificity was 3.03 per cent.
Look at what that means in practice. Ninety-eight per cent of the group scored above 65. Only 34 per cent went on to be diagnosed. The area under the curve was 0.45, which is no better than flipping a coin.
Other work agrees. One study found no difference in scores between adults who were diagnosed and those who were not. Another found more than half of general mental health patients cleared the threshold. Pooled estimates put sensitivity near 75 per cent and specificity near 71 per cent.
| Study | Sensitivity | Specificity |
|---|---|---|
| Original, 2011 | 97% | 100% |
| UK clinic, 2021 | 100% | 3% |
| Pooled estimates | ~75% | ~71% |
| German version, 2024 | 92.5% | 93.6% |
| Revised bands | ~81% | ~81% |
Why the gap is so wide
Three reasons. None of them means the original researchers did anything wrong.
The comparison group changes everything. Separating autistic adults from the general public is easy. Separating them from other people who already suspect they are autistic is very hard. A waiting list self-selects for exactly these traits.
Conditions differed. A clinician was present in the original study. Most people now complete it alone, online.
Other conditions inflate the score. ADHD, social anxiety, trauma and burnout all push it up. The RAADS-R test cannot separate them.
Common scoring mistakes
Four things go wrong often, particularly with online versions.
Reverse-scored items get missed. Some copies simply add every answer. This inflates scores, sometimes badly.
The lifetime options get flattened. A version offering only agree or disagree has thrown away the whole point of the instrument.
The cutoff gets reported as a verdict. Scoring 66 is not the same as being told you are autistic. It is the weakest possible result above an already disputed line.
Subscales get compared as raw numbers. Language has seven items and social relatedness has thirty-nine. Comparing those totals directly is meaningless.
Frequently asked questions
What is a good score on the RAADS-R test?
There is no good or bad score. It measures traits, not performance. Below 65 indicates a low level of autistic traits. Above 65 indicates more. The autistic group in the original study averaged around 134.
How many questions does the RAADS-R test have?
80 questions, split across four subscales: social relatedness (39), sensory-motor (20), circumscribed interests (14) and language (7).
How long does it take?
15 to 20 minutes. There is no time limit. Taking longer is fine, because several items ask what was true before you were 16.
What is the maximum RAADS-R score?
240. Each of the 80 items scores up to 3 points.
Is a score of 65 autistic?
No. A score of 65 clears the published threshold, nothing more. It is the weakest result above the line, and the line itself is disputed. In one UK clinic study, 98 per cent of a waiting list cleared it and only 34 per cent were diagnosed.
Is the RAADS-R test free?
Yes, for personal, clinical and research use. It is published under a Creative Commons Non-Commercial licence. That permits free use with attribution and prohibits commercial use.
Can the RAADS-R test diagnose autism?
No. It is a screening questionnaire. Diagnosis needs a clinical assessment, developmental history and usually several hours with a professional.
Does a low score rule autism out?
No. The RAADS-R test does not ask about masking. Adults who have camouflaged for years can score lower than their experience suggests. This affects women disproportionately.
Can teenagers take it?
It was validated for adults aged 18 and over. For teenagers, parent-report measures such as the SRS-2 or the adolescent AQ are more appropriate.
How does it compare with the AQ?
The RAADS-R test is longer at 80 items against 50, and it takes a lifetime view where the AQ asks about the present. In clinical comparisons both performed in the poor to fair range. Neither asks about masking.
What is the RAADS-14?
A 14-item short form built from the RAADS-R test for busy psychiatric clinics. Its cutoff is 14. Sensitivity is 97 per cent, specificity 46 to 64 per cent.
Should I show my score to a doctor?
Bring examples instead. No clinician treats a self-reported score as evidence. Concrete incidents from your own life in your highest-scoring areas are what an assessment actually needs.
Research and sources
Ritvo RA, Ritvo ER, Guthrie D, et al. (2011). The Ritvo Autism Asperger Diagnostic Scale-Revised (RAADS-R). Journal of Autism and Developmental Disorders, 41(8), 1076-1089. DOI · PubMed
Jones SL, Johnson M, Alty B, Adamou M (2021). The Effectiveness of RAADS-R as a Screening Tool for Adult ASD Populations. Autism Research and Treatment. The 3 per cent specificity study. Open access. Full text
Eriksson JM, Andersen LM, Bejerot S (2013). RAADS-14 Screen. Molecular Autism. Open access. Full text
German validation of the RAADS-R (RADS-R), 2024. Completed without a clinician present. Open access. Full text
Wigham S, et al. (2019). Psychometric properties of questionnaires and diagnostic measures for autism in adults. Autism, 23(2). DOI
Baron-Cohen S, et al. (2001). The Autism-Spectrum Quotient. JADD, 31(1). DOI · PDF
General reference: Autism spectrum, Autism-Spectrum Quotient, Asperger syndrome, Autistic masking and Sensitivity and specificity on Wikipedia.