Alexithymia Test (TAS-20)

The TAS-20 is a revision of the 26-item Toronto Alexithymia Scale. Michael Bagby, James Parker and Graeme Taylor published it in two papers in the Journal of Psychosomatic Research in 1994. The first described how they chose the items and reported a three-factor structure. The second tested whether the scores behaved as expected.

The word alexithymia comes from Greek roots meaning a lack of words for emotion. The psychotherapists Sifneos and Nemiah coined it after noticing that some patients with psychosomatic illness couldn't describe feelings and tended toward a practical, utilitarian style of thinking. In the second paper, the authors compared the scale with personality tests in university students and with observer ratings of alexithymia in behavioral medicine outpatients.

The 1994 abstracts report good internal consistency and test-retest reliability, a three-factor structure that held in clinical and nonclinical groups, and correlations with personality measures that fit the authors' predictions. We could read only those abstracts, so we don't quote the original sample sizes or coefficients.

Later studies tested the structure at scale. A 2022 meta-analysis pooled 88 samples from 62 studies (69,722 people) and found the original three-factor model held up best. In 34 English-language samples (28,514 people), reliability (omega) was .86 for difficulty identifying feelings, .77 for difficulty describing feelings and .64 for externally oriented thinking. A 2003 review covering 19 countries and 18 translations found the three-factor structure held across languages, but in most non-English cultures the third factor lacked internal reliability.

That third factor is the weak spot. In a 2025 study of 1,453 US adults with chronic pain, the three-factor model fit poorly and most externally oriented thinking items loaded below 0.40. A 2026 paper co-written by two of the scale's developers notes that some researchers think the identifying-feelings subscale reflects general distress. The developers recommend using the total score.

The cut-offs of 51 and 61 are the ones open papers that use the TAS-20 report, and one of those papers attributes them to the two 1994 papers. We could read only the abstracts of those 1994 papers, so we haven't confirmed where the numbers come from. The label "Alexithymia present" is this page's wording. A score at or above 61 is a cut-off, not a diagnosis.

Alexithymia isn't a diagnosis in the DSM-5 or ICD-10, though it overlaps with many psychiatric and physical conditions. The TAS-20 is also self-report, so it relies on how well you can see your own feelings, which is the thing it asks about. One of its authors wrote in 2021 that lay people can easily misread a score and that an adequate assessment needs a clinical interview by a professional familiar with the concept.

The cut-offs are fixed numbers, and we couldn't confirm how they were set. Researchers studying chronic pain have warned that three questions about body sensations could inflate scores for people in pain.

Think of your score as a rough reading of how hard feelings are to name right now. A high score doesn't mean something is wrong with you, and a low one doesn't mean naming feelings is easy. Retaking it after a few weeks shows whether the number holds steady.

If trouble with feelings is straining your relationships, work or health, or you also feel low or anxious, it's worth talking with a therapist or doctor, who can ask questions a score can't. For related skills, see our guides to emotion regulation in DBT and emotional numbness.

More on this topic: Emotion regulation in DBT, Emotional numbness, Depression test (PHQ-9), Anxiety test (GAD-7).

Your answers are scored in your browser and aren't sent or saved anywhere. Close the page and they're gone.

• Bagby, Parker and Taylor, Journal of Psychosomatic Research, 1994, paper I (PubMed): abstract only: the 20-item version, good internal consistency and test-retest reliability, three-factor structure replicated in clinical and nonclinical groups.
• Bagby, Taylor and Parker, Journal of Psychosomatic Research, 1994, paper II (PubMed): abstract only: convergent, discriminant and concurrent validity in university students and behavioral medicine outpatients.
• Schroeders, Kubera and Gnambs, Assessment, 2022 (PubMed Central): open-access full text (read through Europe PMC): 88 samples, 62 studies, 69,722 people; omega values for English samples; the origin of the word alexithymia; alexithymia is not in the DSM-5 or ICD-10.
• Taylor, Bagby and Parker, Journal of Psychosomatic Research, 2003 (PubMed): abstract only: 18 languages, 19 countries, three-factor structure generalizes, third factor lacks reliability in most non-English cultures.
• Bagby, Parker and Taylor, Journal of Psychosomatic Research, 2020 (PubMed): abstract only: the authors' own 25-year review; it reaches a positive verdict, so it isn't independent.
• Aaron and colleagues, Pain Reports, 2025 (PubMed Central): open-access full text: prints all 20 items with the reverse-scored ones marked and their factors; 1,453 adults with chronic pain; poor fit of the three-factor model; concern about body-focused items. US online sample, so it says little about other groups.
• Mortezaei, McIntyre, Taylor and Bagby, Behavioral Sciences, 2026 (PubMed Central): open-access full text: the debate over whether the identifying-feelings subscale reflects distress; developers recommend the total score. Co-written by two of the scale's developers.
• Ścigała and colleagues, Frontiers in Psychology, 2020 (PubMed Central): open-access full text of a Polish-language study: the cut-offs of 51 and 61 attributed to the 1994 papers, and the 5-point scale with items 4, 5, 10, 18 and 19 reverse-scored. Secondary source.
• UNSW Sydney outcome measure sheet for the TAS-20: a university summary: which items belong to each factor, the scoring and the cut-offs. Secondary source, no derivation of the cut-offs.
• The Transmitter, June 2021: copyright claim prompts retraction of a study on alexithymia in autism: a news report quoting a letter from one of the authors, who said lay people can misread scores and that assessment needs a clinical interview. Used only for that caution.