Score each of the seven GAD-7 items from 0 ("not at all") to 3 ("nearly every day") and add them for a total between 0 and 21. The original validation suggested 5, 10 and 15 as thresholds for mild, moderate and severe anxiety, with 10 or above the usual cut point for probable generalised anxiety disorder.
That is the quick answer. The part worth knowing is that the severity bands almost every website reproduces were never validated as bands. The authors proposed them by analogy, and said so.
What is the GAD-7 and what does it measure?
The GAD-7 is a seven-item self-report scale asking how often, over the last two weeks, a person has been bothered by feeling nervous or on edge, being unable to stop or control worrying, worrying too much about different things, trouble relaxing, restlessness, irritability, and feeling afraid as if something awful might happen.
It was developed by Robert Spitzer, Janet Williams, Kurt Kroenke and colleagues and published in Archives of Internal Medicine in 2006. The official form states: "No permission required to reproduce, translate, display or distribute." The GAD-7 is free.
Two things it is not: a diagnostic instrument, or a general anxiety measure. It was built to assess the severity of generalised anxiety disorder specifically.
How do you score the GAD-7?
Each item uses the same four response options:
- Not at all: 0
- Several days: 1
- More than half the days: 2
- Nearly every day: 3
Add the seven item scores for a total from 0 to 21. The rating period is the past two weeks, which matters more than it looks. Internal consistency in the original study was excellent (Cronbach's alpha of 0.92) and test-retest reliability good (intraclass correlation of 0.83), so a change in score is more likely signal than measurement noise.
Does the eighth question count towards the total?
No. Most printed versions carry a final question: "If you checked off any problems, how difficult have these problems made it for you to do your work, take care of things at home, or get along with other people?", answered as not difficult at all, somewhat difficult, very difficult or extremely difficult.
That item is not part of the 0 to 21 total. It is a functional impairment probe, and the most commonly mis-scored part of the instrument. It is also the most clinically useful item on the page: two clients scoring 12 who answer "not difficult at all" and "extremely difficult" are not presenting with the same problem.
What do GAD-7 score ranges mean?
The conventional bands are:
- 0 to 4: minimal anxiety
- 5 to 9: mild
- 10 to 14: moderate
- 15 to 21: severe
Where the 5, 10 and 15 bands came from
Here is the part that gets lost. The 2006 paper does not report a validation of these severity levels. It says cut points of 5, 10 and 15 "might be interpreted as representing mild, moderate, and severe levels of anxiety on the GAD-7, similar to levels of depression on the PHQ-9". They were proposed by analogy to the depression scale the same team had published, not derived from anxiety outcome data.
Only the cut point of 10 was tested against a criterion standard. Treat 5 and 15 as conventions for describing a score, not validated clinical boundaries, and never as the sole basis for a treatment decision. The same caution applies to the PHQ-9 severity bands.
Is a GAD-7 of 10 or more a diagnosis of GAD?
No, and the timeframe is the reason. The GAD-7 asks about the past two weeks. A DSM diagnosis of generalised anxiety disorder requires excessive anxiety and worry occurring more days than not for at least six months. The screener cannot see that.
The numbers make the point. In the original study of 965 interviewed primary care patients, 23% scored 10 or above, while the diagnostic interview found generalised anxiety disorder in 9%. At a cut point of 10, sensitivity was 89% and specificity 82%, so most people who screen positive do not have the disorder. Raising the cut point to 15 brought the positive rate to 9%, closer to true prevalence, but sensitivity fell to 48%. Mean scores overlapped: 14.4 with the disorder, 4.9 without.
A score is a prompt for assessment, not a substitute for one.
Should the cut point be 8 or 10?
Both are in active use, for different jobs.
Ten is the cut point validated in the original paper for generalised anxiety disorder. Eight is the threshold used across UK services: the NHS Talking Therapies manual states that for the GAD-7 caseness "is a score of 8 or above", and uses it to decide whether someone is a clinical case at assessment and recovered at discharge.
The lower threshold reflects a different purpose. Kroenke and colleagues (2007) found the GAD-7 also detects panic disorder, social anxiety disorder and post-traumatic stress disorder reasonably well (area under the curve of 0.80 to 0.91 across all four), and a lower cut point catches more of them. Screening broadly for any anxiety disorder, 8 is defensible; asking specifically about generalised anxiety disorder, 10 is the validated answer. Record which one you used.
What is a normal GAD-7 score?
Low. A German general population survey of 5,030 adults, Löwe and colleagues (2008), found a mean of 3.2 for women and 2.7 for men. Roughly 5% scored 10 or above, and 1% scored 15 or above.
Worth quoting to clients who assume a score of 6 means something is seriously wrong. It also explains why comparing a client's score with a population mean is less informative than comparing it with their own score last month.
How much change on the GAD-7 counts as real improvement?
Two independent answers converge on four points.
Toussaint and colleagues (2020) estimated the minimal clinically important difference at 4 points. The NHS Talking Therapies manual sets the reliable change index for the GAD-7 at 4 or more. Two different derivations, the same number.
One caveat: the Toussaint sample came from a trial of chronic depression, not an anxiety trial, and the authors recommend confirmation in anxiety populations. Treat four points as a working threshold rather than a settled fact. In practice, a drop from 16 to 13 is movement but not reliable change; 16 to 11 is reliable change but still above the caseness threshold of 8.
What does the GAD-7 miss?
This is the most consequential limitation and the least discussed. The NHS manual, which uses the GAD-7 as its default anxiety measure, lists the symptoms it does not cover: agoraphobia, body dysmorphic disorder, health anxiety, obsessions and compulsions, panic attacks and the fear of them, intrusive memories and avoidance of trauma reminders, and fear or avoidance of social situations.
So a client with severe social anxiety or panic disorder can score below threshold while remaining significantly unwell. The manual's guidance is to add a disorder-specific measure when one of those conditions is the main focus of treatment, so clinicians do not discharge people whose defining symptoms the scale never asked about.
The GAD-7 also asks nothing about suicidal thoughts. If risk indicators appear elsewhere in the session, document the suicide risk assessment separately.
Does Medicare require a GAD-7 in Australia?
Not the GAD-7 specifically, but an outcome measure is required on the referrer's side. Under Better Access, the MBS descriptor for the GP Mental Health Treatment Plan requires an assessment process including "the administration of an outcome measurement tool (except if considered clinically inappropriate)", and the review requires "re-administration of the outcome measurement tool used in the assessment stage".
Two practical consequences. First, the choice of tool "is at the clinical discretion of the practitioner", but the only examples the descriptor names are the K10 and the DASS 21. The GAD-7 is permitted, it is simply not a named example. Second, the review must re-administer the same tool used at assessment, so if the plan was written on a K10, a GAD-7 at review does not satisfy that requirement on its own.
For psychologists, the psychological therapy and focussed psychological strategies items impose no outcome-measure requirement at all. They do require a written report to the referring practitioner at the end of a course of treatment covering "assessments carried out, treatment provided and recommendations on future management". A GAD-7 you administered belongs in that report, and in the letter back to the referring GP. More on the pathway in our guide to mental health treatment plan referrals.
How should you record a GAD-7 in clinical notes?
An entry that still makes sense at review, or to a colleague reading the file, records:
- Date administered and the rating period (past two weeks)
- The total out of 21, the severity label, and the cut point you applied
- The answer to the unscored difficulty question
- Any single item that stood out, such as item 7 endorsed "nearly every day"
- The comparison point, and whether the change reaches four points
- Any disorder-specific measure added alongside it, and why
Scoring by hand is where small errors creep in: the difficulty item added to the total, a band applied from the wrong cut point, last month's score misremembered. PractaLuma is AI-native practice management software for Australian mental-health practices, and its standardised assessments let you send, score and track the GAD-7, PHQ-9, K10, DASS-21 and other measures so scores sit in the client record and change is visible over time. Results flow into the rest of the clinical record and into anxiety treatment plan goals rather than a separate spreadsheet.
Frequently asked questions
What does a GAD-7 score of 12 mean? It sits in the conventional moderate band (10 to 14), above the usual cut point of 10, so it warrants a clinical assessment. It is not a diagnosis, and says nothing about symptom duration.
Is the GAD-7 free to use? Yes. The form states no permission is required to reproduce, translate, display or distribute it. It was developed with an educational grant from Pfizer Inc.
How often should the GAD-7 be repeated? There is no fixed interval. UK services administer it every session; in Australian private practice, intake, plan review and discharge is common. Use the same cut point and version each time.
What is the GAD-2? The first two items, used as an ultra-brief screen with a cut point of 3. It performed comparably to the full scale in the 2007 primary care study, but gives no severity information, so it is a triage step rather than a monitoring measure.
This guide summarises published scoring conventions for clinicians. It is not a substitute for clinical judgement, current diagnostic criteria, or your organisation's reporting requirements.