The Kessler K10 in one page: ten items on non-specific psychological distress, 0-40 or 10-50 scoring, the Andrews and Slade bands, license and how to cite it.
Ten questions on how often anxiety and depression symptoms occurred over the past 30 days, designed for population health surveys. Its six-item short form is the K6. Free to use with attribution, with official versions for Spain and for the Americas.
By Víctor Ciudad-Fernández, PhD · Updated: October 6, 2026 · Leer esta ficha en español · Instruments
The K10 measures non-specific psychological distress: how often, over the past month, a person has had anxiety and depression symptoms such as nervousness, restlessness, hopelessness, sadness or the sense that everything is an effort. It does not target a specific disorder; it captures how much emotional distress there is, which is exactly what matters for monitoring a population’s mental health.
Kessler and colleagues developed it for the redesigned US National Health Interview Survey (NHIS), building the questions with item response theory models from a larger pool. The six-item form, the K6, is contained within it. Both strongly discriminate between community cases and non-cases of DSM-IV disorders.
It is used in government health surveys in the US and Canada, in the WHO World Mental Health Surveys and, in its K10 form, in Australia’s national mental health survey; Andrews and Slade (2001) also consider it suitable for clinical practice. It works as a screen and as an outcome measure, not as a diagnosis.
Each question has five options by frequency over the past 30 days, from none of the time to all of the time. Under the official rules, each item is coded from 0 (none of the time) to 4 (all of the time) and summed, for a total of 0 to 40. A practical detail: the options are presented from most to least frequent, so they must be recoded so that 0 means no distress.
Australia uses a different coding, 1 to 5 per item, which gives a total of 10 to 50: that is the coding of the Andrews and Slade bands below. Before comparing your data with any reference, check which coding it uses. The official forms (K10+) add questions on persistence and impairment that are not needed to score the scale. The authors accept changing the 30-day frame to the last 7 days if you need to measure change during a brief intervention.
The K10 yields a single distress score. In the original study, the scales showed good precision in the upper tail of the population distribution (90th to 99th percentiles), which is where the cases worth detecting sit, and consistent psychometric properties across major sociodemographic subgroups.
In Spanish older adults, exploratory factor analysis found a single factor explaining 67.5% of the variance, and the confirmatory analysis supported the one-factor model: χ²(35) = 125.08 (WLSMV), CFI = .951, TLI = .938, RMSEA = .066 [.052, .081].
The most cited K10 bands are those of Andrews and Slade (2001), derived from data of the Australian National Survey of Mental Health and Wellbeing. They use the Australian 10 to 50 coding.
| Score (10-50) | Psychological distress | Source |
|---|---|---|
| 10-15 | Low | Andrews & Slade (2001) |
| 16-21 | Moderate | Andrews & Slade (2001) |
| 22-29 | High | Andrews & Slade (2001) |
| 30-50 | Very high | Andrews & Slade (2001) |
If you code 0 to 4 (0 to 40), add 10 to your score before using these bands. For the K6 (0 to 24), the authors give 13 or more as the cut-off for estimating the prevalence of serious mental illness, and warn that it is optimal only in populations with a prevalence similar to that of the US.
A screening questionnaire is not a diagnosis: a high score means a clinical assessment is warranted, not that a disorder is present.
The official Harvard National Comorbidity Survey page offers two Spanish translations of the K10 and K6, one for Spain and one for the Americas, under the same terms of use as the original. In Spain, Sánchez-Moreno, Gallardo-Peralta and Gálvez-Nieto (2025) validated it in 887 people aged 65 or over: one-factor structure, very high reliability and convergent validity with geriatric depression, which makes it suitable for screening distress in older adults.
For general-population comparisons, Lehmann et al. (2023) published age- and sex-specific normative values for seven European countries, Spain included, on the 0 to 40 metric. Its properties have also been studied in Latin American countries; if your sample is from there, use the translation for the Americas and look for the validation in your population.
License: Free to use. According to its official page, no formal permission is needed: they ask you to cite Kessler et al. (2003) and include the copyright (“Copyright © Ronald C. Kessler, PhD. All rights reserved.”), and they appreciate receiving citations of publications that use it. Translations are posted for unrestricted use with the same requirement.
The self-administered and interviewer forms, in English and in the available translations, are downloaded from the K6 and K10 page of the Harvard National Comorbidity Survey, together with the scoring rules.
The official page asks you to cite Kessler et al. (2003). Add the development paper and any validation you use:
@article{kessler2003screening,
author = {Kessler, Ronald C. and Barker, Peggy R. and Colpe, Lisa J. and Epstein, Joan F. and Gfroerer, Joseph C. and Hiripi, Eva and Howes, Mary J. and Normand, Sharon-Lise T. and Manderscheid, Ronald W. and Walters, Ellen E. and Zaslavsky, Alan M.},
title = {Screening for serious mental illness in the general population},
journal = {Archives of General Psychiatry},
year = {2003},
volume = {60},
number = {2},
pages = {184--189},
doi = {10.1001/archpsyc.60.2.184}
}As the authors themselves note, a cut-off is only optimal in a population with the prevalence it was calibrated in. If you compare the K10 with a clinical interview in your sample, compute the sensitivity and specificity of your cut-off with the sensitivity and specificity calculator, which also gives you predictive values and likelihood ratios.
The K6 is six of the ten items and discriminates almost as well; it was built for surveys where every question counts. The K10 gives a little more information and is the one with the Andrews and Slade bands. If you will compare with Australian data, use the K10 with 1 to 5 coding.
Yes, if you convert: the 1 to 5 coding adds 10 points to the 0 to 4 coding (one point per item). Subtract 10 from a 10-50 score to get the 0-40 metric, or add 10 to go the other way.
This fact sheet is an academic reference for information purposes: it does not reproduce any questionnaire item. Using the instrument in your research or practice may require citing its authors or asking for permission: see “How to get the instrument”.