EAT-26: Eating Attitudes Test

What this assessment is

The Eating Attitudes Test, 26-item version, or EAT-26, is a self-report screen for eating-disorder symptoms and concerns.

David Garner, Marion Olmsted, Yvonne Bohr, and Paul Garfinkel developed it in 1982 as a shorter form of the original 40-item test. An Australian eating-disorders research hub identifies the standard version and original development paper.

The EAT-26 identifies reasons for further evaluation. It does not diagnose an eating disorder.

That distinction is central.

Who it was designed for

The EAT-26 was developed from research involving people with anorexia nervosa and comparison groups. It has since been used with adolescents and adults in clinical, school, community, and research settings.

The meaning of a result can change with age, gender, language, culture, medical history, and the reason for screening. Evidence from one group does not guarantee the same accuracy in another.

Use a validated version for the population and language being assessed.

Do not assume equivalence.

How it is administered

The EAT-26 is self-completed. Its core section has 26 statements with six response choices.

The standard screening package also asks about specified behaviors and recent weight change. Those responses are not added to the core 26-item total. They are considered separately when deciding whether further evaluation is appropriate.

The total is separate.

Use the complete form and scoring system. A shortened set of questions or changed response scale is not the EAT-26.

What it measures

The core total covers eating attitudes and behaviors associated with dieting, food preoccupation, perceived social pressure, and oral control.

These domains can identify concern, but they do not capture every eating-disorder presentation.

No short screen can.

The SCOFF is a shorter five-question suspicion screen. It does not use the EAT-26 response scale, total, or referral system.

How scoring works

Each core response scores from 0 to 3 under the standard key. One item is keyed in the opposite direction. Add the 26 values for a total from 0 to 78.

Higher totals indicate more reported eating-disorder symptoms and concerns.

Higher does not mean diagnosed.

Use the official scoring key. Do not assign points according to the visual order of every response, because the reverse-keyed item would be scored incorrectly.

The behavioral and weight-related referral criteria remain separate. Do not add them to the 0-to-78 total.

Keep them separate.

How to interpret an EAT-26 score

The developer's interpretation guidance uses a core total of 20 or higher as one reason for a professional evaluation.

It also identifies specified behavioral responses and low body weight as separate referral criteria. Meeting any one criterion can support further evaluation, even when the core total is below 20.

The threshold is not a diagnostic boundary. A score of 20 does not prove an eating disorder, and a score of 19 does not rule one out.

It is a referral signal.

Do not divide the total into unsupported mild, moderate, or severe bands. The standard interpretation system does not define those categories.

What the score cannot tell you

The EAT-26 cannot establish a diagnosis, identify a specific eating disorder, or determine whether someone is medically stable.

It cannot fully assess current intake, compensatory behavior, weight trajectory, physical complications, or the effect on daily life.

A lower total can occur despite serious concern. Symptoms may fall outside the core domains, and people may underreport experiences because of shame, fear, or limited insight.

Look beyond the total.

Clinical urgency depends on the whole picture, not the core score.

Evidence and limitations

The EAT-26 reduced the original test from 40 items while retaining a strong relationship with the longer total. Its shorter format helped it become widely used in screening and research.

Wide use does not create one accuracy estimate for every population. Translations, settings, sampling methods, and the mix of eating-disorder presentations differ across studies.

Variation is expected.

Interpret the result with the separate referral criteria, current health, and a full assessment. Do not use the total as a diagnosis or a measure of medical risk.