How to Calculate a Global Assessment of Functioning (GAF) Score

The Global Assessment of Functioning (GAF) score is a single number between 1 and 100 that a clinician assigns to capture how well a person is doing overall, considering both psychological symptoms and day-to-day functioning. Calculating it involves matching the person’s current presentation to one of ten descriptive brackets on the scale, then refining within that bracket. The process sounds straightforward, but the details matter enormously, and getting it right requires more judgment than most rating tools in psychiatry.

What the GAF Scale Actually Measures

The GAF was included as Axis V in the DSM-IV multiaxial diagnostic system, where it served as a quick snapshot of a patient’s overall mental health status at a given point in time.¹ Its purpose was to give clinicians a way to supplement diagnostic labels with a number reflecting how impaired or well-functioning someone actually was. A person with major depression could be scoring a 70 (mild symptoms, generally functioning well) or a 30 (behavior considerably influenced by delusions or hallucinations, or serious impairment in communication or judgment). The diagnosis alone doesn’t tell you that; the GAF was designed to fill the gap.

The scale runs from 1 (persistently dangerous, unable to maintain minimal personal hygiene) to 100 (superior functioning in a wide range of activities, no symptoms). It captures two separate dimensions in a single number: symptom severity and functional impairment. This dual nature is both the scale’s strength and its main source of confusion when you sit down to score it.

The Step-by-Step Scoring Process

Scoring involves two main stages. First, you identify the correct 10-point bracket. Then you narrow down within that bracket to a specific number.

Each 10-point range on the GAF has a written description covering both symptoms and functioning. The 41–50 bracket, for instance, describes serious symptoms like suicidal ideation or severe obsessional rituals, and serious functional impairment such as having no friends or being unable to keep a job. The 61–70 bracket describes mild symptoms like depressed mood and mild insomnia, and some difficulty in social or occupational functioning but generally functioning pretty well. You read through the descriptors and find the bracket that best matches the person you’re evaluating.

The critical rule is to use the lower of the two dimensions. If someone’s symptoms place them in the 51–60 range (moderate symptoms) but their functioning places them in the 31–40 range (major impairment in several areas), the GAF score comes from the 31–40 bracket. Research confirms this approach works reasonably well as a global indicator, and when symptoms and functioning do diverge substantially, it happens in roughly one in ten cases, with functioning usually being the more impaired of the two.²

Narrowing Down Within a Bracket

Once you’ve identified the right 10-point range, you need to pick a specific number. This is where things get subjective. The official guidelines from the DSM-IV-TR, the Veterans Administration, and Norwegian clinical guidelines all give some instructions for scoring within the brackets, but those instructions are limited.³ In practice, clinicians tend to gravitate toward the round numbers at the top or middle of each bracket. If the bracket is 41–50, you’ll see a lot of scores at 45 and 50, fewer at 42 or 47.

A useful approach when deciding between, say, 43 and 48 is to ask yourself whether the person fits the bracket description fully and cleanly, or whether they’re on the edge. Someone who matches every aspect of the bracket description and has few features from the bracket above would sit in the lower half. Someone who matches the bracket but also shows hints of the higher bracket’s descriptors would land toward the top. Patients who are scored within the same 10-point interval should be relatively similar in their overall functioning, though the concept of “functioning” has many facets, and when detailed information is lacking, clinicians tend to pick middle-of-the-range scores.³

The GAF Brackets at a Glance

While the full DSM-IV descriptions are more nuanced, here is a simplified summary of what each range broadly represents:

91–100: Superior functioning across the board, no symptoms.
81–90: Absent or minimal symptoms, good functioning in all areas.
71–80: Symptoms are transient and expected reactions to stressors; slight impairment in functioning.
61–70: Mild symptoms or mild difficulty in social or work functioning, but generally doing well.
51–60: Moderate symptoms or moderate difficulty in functioning.
41–50: Serious symptoms or serious impairment in social or occupational functioning.
31–40: Major impairment in several areas like work, family relations, judgment, thinking, or mood.
21–30: Behavior considerably influenced by delusions or hallucinations, or serious impairment in communication or judgment.
11–20: Some danger of hurting self or others, or occasionally fails to maintain minimal personal hygiene, or gross impairment in communication.
1–10: Persistent danger of severely hurting self or others, or persistent inability to maintain minimal personal hygiene, or serious suicidal act with clear expectation of death.

The descriptors in each bracket contain “or” language throughout, meaning a person doesn’t need to match every feature listed. Matching either the symptom description or the functioning description is enough to place them in that bracket.

Why Two Clinicians Often Disagree

One of the most persistent criticisms of the GAF is that different raters can arrive at meaningfully different scores for the same patient. In routine clinical settings, inter-rater reliability between clinicians and trained researchers has been found to be low, with agreement coefficients ranging from about 0.39 to 0.59.⁴ When two trained researchers rate the same patient, agreement is much higher, often above 0.80.⁴ The gap suggests the problem isn’t the scale itself so much as how it’s used in the real world.

In one study of outpatients with depressive disorders, the average GAF score assigned by clinicians was about 55, while a test nurse evaluating the same patients gave an average of about 58, and the two sets of scores correlated weakly.⁵ A three-point gap might sound small, but when you’re making decisions about treatment intensity or disability benefits, that margin can matter.

Interestingly, when clinical psychiatric staff who regularly assess first-time patients use the GAF, reliability tends to be satisfactory, with the key factors being the rater’s attitude toward the scale and their general motivation to use measurement tools in clinical work.⁶ Clinicians who take the scale seriously and use it carefully get more consistent results than those who treat it as an afterthought.

Training Makes a Real Difference

A study conducted in Uganda looked at what happens when psychiatric clinical officers receive a brief, focused training session on the GAF. Before training, the agreement between clinical officers and psychiatrists was modest. After a short training program, agreement improved substantially, jumping from coefficients in the 0.48–0.59 range to 0.60–0.83 range.⁷ The takeaway is practical: if you’re going to use the GAF, even a brief training session on scoring conventions can dramatically improve how consistent your scores are with those of other clinicians.

The most common scoring errors come from a few habits. Some raters anchor too heavily on the diagnosis rather than evaluating actual functioning. A schizophrenia diagnosis gets a lower score than an anxiety diagnosis regardless of how the individual is actually doing. Other raters conflate symptoms with functioning, giving high scores when symptoms are controlled even if the person still can’t hold a job. The scale explicitly asks you to rate the lower of the two, and being aware of this rule is the single biggest step toward scoring accurately.

Where the GAF Score Still Matters

Despite being officially removed from the DSM-5, the GAF hasn’t disappeared from clinical and legal life. The Veterans Affairs system has used GAF scores extensively as one input in mental health outcome monitoring, though researchers have cautioned that the scale should not be used to compare outcomes across different facilities due to uncontrolled variation in how it’s administered.⁸ VA disability ratings for mental health conditions have historically referenced GAF scores, and many veterans seeking benefits still encounter them in their records. The VA rating schedule has since moved away from requiring a GAF score, but older evaluations and appeal documents frequently include them.

In disability evaluations more broadly, psychiatrists perform assessments in the context of Social Security Disability Insurance applications, workers’ compensation claims, and private disability insurance claims.⁹ A GAF score can carry weight in these proceedings, particularly when older clinical records are being reviewed. If you’re a clinician documenting a GAF score that could end up in a legal context, precision and consistency in your scoring method matter more than usual.

The DSM-5 Shift and the WHODAS 2.0

When the DSM-5 was published in 2013, it dropped the multiaxial system entirely and eliminated the GAF as the endorsed assessment tool for functioning.¹⁰ In its place, the DSM-5 recommended the World Health Organization Disability Assessment Schedule 2.0 (WHODAS 2.0).¹¹

The main reasoning behind this change was the GAF’s known reliability problems in routine use and the way it blended symptoms with functioning in a single number. The WHODAS 2.0 measures disability across six life domains (cognition, mobility, self-care, getting along with people, life activities, and participation in society) and can be completed by the patient themselves, removing some of the clinician-variability problem. It also intentionally separates functioning from symptom severity, which the GAF’s “use the lower score” rule never fully did.

That said, many clinicians and institutions haven’t fully transitioned. Research papers still report GAF scores, older clinical records still contain them, and some international health systems continue to use the GAF routinely. If you encounter a GAF score in a chart, understanding what it means and how it was derived remains practically important.

GAF Scores as Predictors

Beyond serving as a snapshot, GAF scores carry some prognostic value. In a large study of people diagnosed with schizophrenia for the first time, lower GAF functioning scores at the time of diagnosis were associated with a higher risk of later hospitalization, following a dose-response pattern: the lower the score, the higher the risk. Among males, those with GAF functioning scores between 1 and 20 had roughly double the hospitalization risk compared to those scoring 61–100.¹² The relationship was less clear in females in that study, a finding the researchers didn’t fully explain.

Similarly, in patients with bipolar disorder, a lower GAF score at the time of hospital discharge was a significant predictor of psychiatric readmission.¹³ Clinicians sometimes treat the GAF as a formality to fill out in the chart, but these findings suggest it’s picking up on something real about who is likely to deteriorate.

Related Scales Worth Knowing About

The GAF didn’t exist in a vacuum, and a few related scales come up frequently in the same contexts. The Social and Occupational Functioning Assessment Scale (SOFAS) is structured almost identically to the GAF but deliberately excludes symptoms from the score. It rates only how well a person is functioning socially and at work, without penalizing for symptom severity alone. Research has found that GAF and SOFAS total scores are practically exchangeable, though the SOFAS can sometimes reveal slight functional impairment even in patients who are symptom-free.¹⁴

The Global Assessment of Relational Functioning (GARF) takes a different approach entirely, rating the quality of a person’s relational context rather than individual functioning. It looks at three areas: joint problem-solving, organizational structure within relationships, and emotional climate.¹⁵ The GARF was designed as a companion to the GAF, alerting clinicians to evaluate not just the individual but the relational system they live in. It was included as a proposed measure in the DSM-IV appendix and investigated alongside the GAF in reliability studies.¹⁶

The Children’s Version

The GAF was developed for adults, but a parallel tool exists for younger populations. The Children’s Global Assessment Scale (CGAS) is adapted from the adult version and is designed for rating general functioning in children and adolescents aged 4 to 16.¹⁷ It follows the same 1–100 structure with descriptive brackets, but the anchor points are tailored to developmentally appropriate expectations. A child who can’t function at school and has no peer relationships is rated differently from an adult who can’t hold a job and has no friends, because the developmental context is different. The CGAS has been used in research examining how impairment criteria affect diagnostic rates in children of depressed and non-depressed parents.¹⁸

Common Mistakes When Scoring

If you’re learning to use the GAF or reviewing scores in a clinical record, a few recurring errors are worth flagging.

The most frequent mistake is rating based on potential rather than current presentation. The GAF asks about functioning right now, during the current period being assessed. A patient who was working full-time six months ago but is currently unable to leave the house should be scored based on their current state, not their baseline when well. Some clinicians instinctively rate toward where they believe the person “should” be, which inflates scores.

Another common error is ignoring the “or” structure of the descriptors. Each bracket contains both symptom anchors and functioning anchors connected by “or.” You don’t need to match both. A person with severe symptoms but adequate functioning still gets placed according to their symptom severity if that’s the lower of the two. Forgetting this leads to artificially high scores for people who are managing to hold their lives together despite profound distress.

Finally, there’s a tendency to avoid the extremes of the scale. Scores above 90 or below 10 are rare in practice, partly because they genuinely apply to fewer people, but also because clinicians feel uncomfortable with the implied certainty. If someone truly has no symptoms and is functioning across all areas at a level others admire, a score above 90 is appropriate. The top and bottom of the scale exist for a reason, and treating the functional range as 30–70 truncates the tool’s sensitivity.

When a GAF Score Shows Up in Your Records

If you’re a patient who has found a GAF score in your own medical records or in a disability evaluation report, keep a few things in mind. The number reflects one clinician’s judgment at one point in time. It is not a permanent label, and it can change dramatically as your condition improves or worsens. Two different clinicians evaluating you on the same day might produce scores that differ by 5 to 10 points or more, which is a known limitation of the tool rather than a sign that one of them got it wrong.

In legal and benefits contexts, a single GAF score rarely decides anything on its own. Adjudicators typically look at the pattern of scores over time, the consistency between the GAF number and the narrative description of functioning in the clinical note, and how the score fits with other evidence in the file. A GAF of 45 paired with a note describing someone who works part-time, socializes regularly, and manages their own finances raises questions, because those details don’t match a score in the “serious impairment” range. Internal consistency between the number and the clinical narrative is what gives a GAF score its real weight in a record.