Interactive Interpreter

EAT-26 Scoring: Eating Attitudes Test (26-Item), Cutoff 20

The EAT-26 (Eating Attitudes Test-26) is a 26-item self-report questionnaire that screens for eating disorder risk by measuring attitudes and behaviours around dieting, food preoccupation and control of eating. Each item scores 0 to 3 for a total of 0 to 78, and a total of 20 or more is the referral threshold for a clinical interview.

For professional reference, research, and education only. Not for diagnosis or medical advice.

EAT-26 Score Interpreter

  • Enter a total score to see its range.

  • 26 items. Items 1 to 25: Always 3, Usually 2, Often 1, other answers 0. Item 26 is reverse-scored (Never 3). Sum for a total of 0 to 78.

    Total scoreInterpretation
    20+At or above the referral thresholdA total of 20 or more: refer for interview by a qualified professional. This is a screen, not a diagnosis.
    0–19Below the total-score thresholdA total under 20 does not rule out an eating disorder. Refer anyway if the client reports any of the listed eating disorder behaviours or is underweight for age and sex.

    Referral threshold of 20 or more per the EAT-26 scoring guide (Garner, Eating Attitudes, LLC). The behavioural questions and underweight status are separate referral criteria.

The EAT-26 (Eating Attitudes Test-26) is a 26-item self-report questionnaire that screens for eating disorder risk; a total of 20 or more out of 78 means the client should be interviewed by a qualified professional 2.

What does the EAT-26 measure?

The EAT-26 measures attitudes and behaviours linked to eating disorders. Garner and colleagues derived it from the 40-item Eating Attitudes Test through factor analysis, and the two versions correlate at r 0.98 1.

  • Three subscales. Dieting, bulimia and food preoccupation, and oral control. The subscales related to bulimia, weight, body image and psychological symptoms in the original sample 1.
  • Behavioural questions. Separate questions ask about specific eating disorder behaviours and weight loss. They feed the referral decision alongside the total and body mass index 2.

It was first proposed as a measure of the symptoms of anorexia nervosa 1 and is now used more broadly to screen for eating disorder risk.

How do you give the EAT-26?

  • Format. 26 statements answered on a six-option frequency scale from Always to Never, plus the behavioural questions, height and weight 2.
  • Who completes it. The client, on paper or on screen. It is designed for use by mental health professionals and others gathering referral information, in settings from schools to clinics.
  • Terms of use. Eating Attitudes, LLC holds the copyright. Fees and royalties are waived for individual users, reproduction needs a permission letter, and commercial web platforms that charge user fees need a licensing agreement 7.
  • Approach. Ask about eating and weight without judgment, and use weight-neutral language when talking through the answers.

How do you score and interpret the EAT-26?

  1. Score items 1 to 25: Always 3, Usually 2, Often 1, any other answer 0 2.
  2. Score item 26 in the opposite direction, so Never scores 3 2.
  3. Add the 26 scores for a total of 0 to 78.
  4. Refer for a clinical interview if the total is 20 or more 2.
  5. Also refer if the client reports any of the listed eating disorder behaviours, or if their body mass index is in the underweight range for their age and sex, whatever the total 2.
  6. For missing answers, the scoring guide suggests estimating up to one missing value per subscale from the median of that subscale's other items 2.

Psychometric properties

  • Original study. In 160 women with anorexia nervosa and 140 comparison women, the EAT-26 correlated 0.98 with the EAT-40, and its three factors related meaningfully to bulimia, weight, body image and psychological symptoms 1.
  • DSM-IV criteria. In non-clinical women, the EAT had an accuracy rate of at least 90% for separating those with and without an eating disorder 3.
  • Spanish women. Item scores formed a single factor with good internal consistency. Specificity was good, but sensitivity was insufficient in a community sample and moderate in a clinical comparison 4.
  • Brazilian women. Against a diagnostic interview, sensitivity was 40%, specificity 84% and positive predictive value 14%; Cronbach's α was 0.75 5.
  • Colombian women. A modified, abbreviated Spanish version had Cronbach's α 0.92, and in that validation sample a cutoff of 11 identified every interview-confirmed case (specificity 85.6%) 6. That cutoff applies to the modified version only.

Limitations

  • Screening, not diagnosis. A positive screen calls for a clinical interview.
  • False positives in community samples. It was built to describe anorexia nervosa symptoms, and in non-clinical samples it can flag many people without an eating disorder 3.
  • Missed cases. Sensitivity was low in some community studies 4 5, so the behavioural questions and clinical judgment matter even below 20.
  • Narrower coverage. It centres on restrictive and bulimic features. For binge eating severity use the BES, for ARFID patterns the NIAS, and for probable diagnoses the EDDS.
  • Licensing. Online platforms that charge user fees need a licence from the copyright holder 7.
  • Related measures. The SCOFF is a shorter screen; the EDE-Q tracks severity over time.

For individuals: This page is a clinical reference, not a diagnostic tool. If you have concerns about your own symptoms, bring this information to a qualified mental health professional.

Measuring progress with clients?

  • HiBoop sends validated measures such as the PHQ-9 and GAD-7 by email or QR code, scores them on submission and charts each client's results over time.

    Start tracking scores

Related

FAQ

  • How is the EAT-26 scored?

    Items 1 to 25 score 3 for Always, 2 for Usually, 1 for Often and 0 for any other answer. Item 26 is scored in the opposite direction, so Never scores 3. The 26 item scores are added for a total from 0 to 78.

  • What is a high EAT-26 score?

    A total of 20 or more is the referral threshold: the client should be interviewed by a qualified professional. The EAT-26 is a screen, so a high score means further assessment is needed, not that an eating disorder is present.

  • Does a score under 20 rule out an eating disorder?

    No. The scoring guide also asks for a referral when the client reports any of the listed eating disorder behaviours, or is underweight for their age and sex, whatever the total. Studies have also found that the EAT-26 misses some people who have an eating disorder.

  • What does the EAT-26 measure?

    The EAT-26 measures attitudes and behaviours linked to eating disorders. Garner and colleagues derived it in 1982 from the 40-item Eating Attitudes Test, and it has three subscales: dieting, bulimia and food preoccupation, and oral control.

References

    1. 1.Garner DM, Olmsted MP, Bohr Y, Garfinkel PE. The eating attitudes test: psychometric features and clinical correlates. Psychol Med. 1982;12(4):871-878.
    2. 2.Garner DM. EAT-26 scoring and interpretation. Eating Attitudes, LLC.
    3. 3.Mintz LB, O'Halloran MS. The Eating Attitudes Test: validation with DSM-IV eating disorder criteria. J Pers Assess. 2000;74(3):489-503.
    4. 4.Rivas T, Bersabé R, Jiménez M, Berrocal C. The Eating Attitudes Test (EAT-26): reliability and validity in Spanish female samples. Span J Psychol. 2010;13(2):1044-1056.
    1. 5.Nunes MA, Camey S, Olinto MT, Mari JJ. The validity and 4-year test-retest reliability of the Brazilian version of the Eating Attitudes Test-26. Braz J Med Biol Res. 2005;38(11):1655-1662.
    2. 6.Constaín GA, Ricardo Ramírez C, Rodríguez-Gázquez ML, et al. Diagnostic validity and usefulness of the Eating Attitudes Test-26 for the assessment of eating disorders risk in a Colombian female population. Aten Primaria. 2014;46(6):283-289.
    3. 7.Garner DM. EAT-26 permission to reproduce. Eating Attitudes, LLC.

  • EAT-26 © Eating Attitudes, LLC (David M. Garner). Protected under copyright; fees and royalties are waived for individual users, and commercial web platforms that charge user fees must obtain a licensing agreement from the copyright holder.

HiBoop does not offer this scale. This page is reference material.

Start