What is the CARS-2? A guide to the Childhood Autism Rating Scale, Second Edition

The CARS-2, or "Childhood Autism Rating Scale, Second Edition," is a 15-item rating scale that a trained professional completes to help identify autism spectrum disorder and describe how pronounced its features are. It supports a diagnosis but does not by itself make one, and it is not an IQ test. Published by Western Psychological Services (WPS) in 2010, the CARS-2 comes in a Standard Version, a High-Functioning Version, and an unscored questionnaire for parents and caregivers. This article covers what each version does, how the ratings and severity ranges work, who administers the scale, and how it differs from the ADOS-2.

The original Childhood Autism Rating Scale grew out of the TEACCH program in North Carolina, where clinicians needed a brief, structured way to summarize their observations of children referred for evaluation. The second edition, by Eric Schopler, Mary E. Van Bourgondien, Glenna Janette Wellman, and Steven R. Love, was published by WPS in 2010. According to the publisher, the CARS-2 "helps to identify children with autism and determine symptom severity through quantifiable ratings based on direct observation."

The scale is designed for individuals ages 2 years and up and is used in clinics, schools, and intervention programs. It is a "Level C" instrument, meaning WPS restricts its purchase to professionals with advanced training in psychological assessment. The authors are explicit that the CARS-2 is intended for people who have already been referred for evaluation. It was never meant as a screener for the general population.

The kit contains two clinician-completed rating booklets and one questionnaire, each with a distinct role.

• Standard Version (CARS2-ST): Equivalent to the original CARS. The publisher recommends it for children younger than 6 years, and for older individuals who have notable communication difficulties or estimated IQs of 79 or lower.

• High-Functioning Version (CARS2-HF): New in the second edition. It is designed for verbally fluent individuals ages 6 and older with estimated IQs of 80 or higher, whose social and behavioral differences tend to be subtler and easy to miss on the original item set.

• Questionnaire for Parents or Caregivers (CARS2-QPC): An unscored form completed by a parent or caregiver. It gathers information about early development, social and communication skills, repetitive behaviors, play, routines, and sensory interests, and it often serves as the framework for a follow-up interview. It never produces a score on its own; it simply informs the clinician's ratings.

Both rating booklets contain 15 items covering areas such as relating to people, imitation, emotional response, body use, object use, adaptation to change, visual and listening responses, verbal and nonverbal communication, activity level, consistency of intellectual response, and the rater's general impressions. Several items on the High-Functioning Version were reworded to capture subtler behavior. For example, "Imitation" becomes "Social-Emotional Understanding," and "Activity Level" becomes "Thinking/Cognitive Integration Skills."

For each of the 15 items, the clinician assigns a rating from 1 (within normal limits for that age) to 4 (severely abnormal for that age). Midpoint ratings of 1.5, 2.5, and 3.5 are allowed, so there are seven possible values per item. Ratings reflect the frequency, intensity, peculiarity, and duration of the behavior in question rather than a simple yes-or-no judgment. The 15 ratings are summed into a total raw score that can range from 15 to 60.

The manual groups Standard Version raw scores into three severity ranges. Scores below 30 fall in the minimal-to-no symptoms range, scores from 30 to 36.5 fall in the mild-to-moderate symptoms range, and scores of 37 and higher fall in the severe symptoms range. For individuals 13 and older, the manual sets the bands slightly lower, with mild-to-moderate beginning at 28 and severe at 35. The High-Functioning Version uses lower thresholds throughout: 28 to 33.5 for mild-to-moderate symptoms and 34 or higher for severe symptoms. When a total raw score reaches the mild-to-moderate range, the manual advises that autism spectrum disorder should become a strong hypothesis to examine within a comprehensive differential diagnosis, and that further evaluation is warranted.

The booklets also convert the total raw score to a "T-score," a standard score with a mean of 50, and a percentile rank. These comparisons are based on a clinical sample of 1,034 individuals with autism spectrum diagnoses, so they describe symptom levels relative to other people on the spectrum rather than to the general population. As with any test, the numbers carry measurement error; the manual reports a standard error of about 3 T-score points, so a single administration should be read as a range rather than an exact value.

Researchers continue to study where the cutoffs should sit. A 2023 validation study in the Journal of the Korean Academy of Child and Adolescent Psychiatry found that a Standard Version cutoff of 28.5 caught more children on the spectrum in its clinical sample than the conventional 30. Findings like these are one more reason scores feed clinical judgment instead of replacing it.

The CARS-2 can be completed by a range of professionals, including psychologists, physicians, speech-language pathologists, and special educators, provided they have training and experience with autism. Before rating, the clinician gathers information from direct observation, developmental history, prior records, and the parent or caregiver questionnaire, then completes the booklet independently in about 5 to 10 minutes.

The manual is direct on this point: the CARS-2 by itself does not produce a diagnosis. Ratings are one piece of a multifaceted evaluation, and the final diagnostic decision belongs to experienced clinicians who are qualified to make it. In practice, a thorough autism evaluation usually combines the CARS-2 or a similar instrument with a developmental interview, cognitive testing, adaptive behavior measures such as the Vineland-3, and often other rating scales such as the SRS-2, which parents and teachers complete.

Families sometimes hear both names during an evaluation, and the two are structurally different tools. The ADOS-2 (Autism Diagnostic Observation Schedule, Second Edition) is a semi-structured, standardized observation in which the examiner personally administers a series of activities and social presses, then codes the behavior that occurs during that session. The CARS-2 is a rating scale: the clinician synthesizes everything known about the person, including observation, records, and caregiver input, into 15 ratings. One samples behavior directly in a controlled session; the other quantifies the clinician's overall impression from multiple sources.

The two approaches tend to agree. The 2023 Korean validation study reported a correlation of about .86 between CARS2-ST and ADOS-2 total scores, and the CARS-2 manual reports correlations near .8 between its forms and the original ADOS. Many clinics use both, since the observation schedule and the rating scale can catch different information, and each is still only a component of the broader evaluation.

The CARS-2 has earned its wide use, and it also has boundaries worth understanding.

• Strengths: It is brief, inexpensive, and backed by decades of research on the original scale. Using the traditional cutoff of 30, the scale identified autism with a "sensitivity" of .88 (the share of true cases flagged) and a "specificity" of .86 (the share of non-cases correctly passed over), figures a 2025 study in the Journal of Autism and Developmental Disorders describes as good. Total-score reliability is high, about .93 to .96 across the two forms, and the High-Functioning Version extends coverage to people the original CARS often missed.

• Limitations: Its standard scores are based on clinical samples of people with autism diagnoses rather than the general population. Individual item scores carry considerable measurement error, so the total score is the only number that should drive decisions. The best cutoff varies by age, ability level, and population, which is an active research question. And because rating quality depends on the rater's information and experience, the same child can receive different scores from differently informed raters.

One boundary deserves special emphasis: the CARS-2 measures behavior associated with autism, and nothing about intelligence. An autistic person can have an IQ anywhere on the scale, a topic explored in our article on how IQ and autism relate. The CARS-2 asks which version to use based on estimated IQ, so a separate, individually administered cognitive test is often part of the same evaluation.

No online test screens for or detects autism, and no online test replaces an individually administered diagnostic evaluation by a qualified clinician. Online cognitive testing has a different, narrower job: measuring reasoning abilities in adults. The RIOT IQ test, developed by RIOT IQ with psychometrician Dr. Russell T. Warne, measures adult (18+) cognitive ability with 15 subtests across six cognitive indices (verbal reasoning, fluid reasoning, spatial ability, working memory, processing speed, and reaction time), takes about 52 minutes, and reports scores on the familiar mean-100, standard-deviation-15 scale. Used for what it is designed for, it is a rigorous way for adults to learn about their cognitive strengths.

No. The CARS-2 rates behaviors associated with autism spectrum disorder. It does not measure intelligence, and its scores say nothing about a person's cognitive ability.

On the Standard Version, a total raw score of 30 falls at the bottom of the mild-to-moderate symptoms range. It signals that autism should be seriously considered and evaluated further. It is a flag for clinical judgment, and it is never a diagnosis by itself.

Professionals with graduate training in assessment and experience with autism, such as psychologists, physicians, speech-language pathologists, and special educators. WPS restricts purchase to qualified users.

The Standard Version is for children under 6 and for individuals with limited verbal skills or estimated IQs of 79 or lower. The High-Functioning Version is for verbally fluent individuals ages 6 and up with estimated IQs of 80 or higher, and it uses lower severity cutoffs.

Parents complete the unscored CARS2-QPC questionnaire, which informs the clinician's ratings. The 15 scored items are always rated by the professional.

No. Autism is diagnosed by qualified clinicians through a comprehensive evaluation that combines observation, history, and standardized instruments administered in person.

1. Western Psychological Services. (2010). (CARS-2) Childhood Autism Rating Scale, Second Edition. wpspublish.com

2. Ji, S.-I., Park, H., Yoon, S. A., & Hong, S.-B. (2023). A validation study of the CARS-2 compared with the ADOS-2 in the diagnosis of autism spectrum disorder: A suggestion for cutoff scores. Journal of the Korean Academy of Child and Adolescent Psychiatry, 34(1), 45–50. jkacap.org

3. Vaughan, C. A. (2011). Review of Childhood Autism Rating Scale (2nd ed.). Journal of Psychoeducational Assessment, 29(5), 489–493. doi.org

4. Chavers Edgar, T., Elmquist, M., Schabes, C., Mohammed, L., Banasik, A., & Sterling, A. (2025). Assessing autism co-occurrence in fragile X syndrome: Proposing a preliminary CARS-2 cut-off score. Journal of Autism and Developmental Disorders. pmc.ncbi.nlm.nih.gov

5. Texas Statewide Leadership for Autism Training. (n.d.). Childhood Autism Rating Scale, 2nd Edition. txautism.net

Photo by Shixart1985, via Wikimedia Commons (CC BY 2.0)