The Global Assessment of Functioning, or GAF, is a single-number rating from 1 to 100 that clinicians use to summarize how well a person is doing psychologically, socially, and occupationally. It was part of the official psychiatric diagnostic system in the United States for decades, appearing on Axis V of the DSM-IV. Despite its widespread use in clinical settings, disability evaluations, and research across many countries, the GAF has been dogged by concerns about its reliability and what it actually measures, and it was formally dropped from the DSM-5 in 2013. Even so, the score remains deeply embedded in veteran disability claims, court proceedings, and treatment planning around the world.
How the Scale Works
The GAF runs from 1, representing the most severe impairment, to 100, representing superior functioning with no symptoms. The scale is divided into ten ranges of ten points each. A score of 91 to 100 describes someone with no symptoms and excellent functioning across all areas of life. A score of 41 to 50 indicates serious symptoms or serious impairment in social or occupational functioning. Below 20, a person may be in danger of hurting themselves or others or unable to maintain basic personal hygiene.
In practice, clinicians are supposed to pick the ten-point range that best matches the patient, then refine within that range. But the instructions for choosing, say, a 43 versus a 47 are thin. The DSM-IV-TR and various national guidelines offer some direction for landing within a ten-point band, but the finer distinctions are largely left to the clinician’s judgment. As a result, scores tend to cluster around round numbers and mid-decade marks rather than spreading evenly across the scale. When detailed information about a patient is limited, clinicians default to middle-of-the-range estimates, which makes scores within the same bracket less meaningful than they appear.
1PubMed Central. Guidelines for rating Global Assessment of Functioning (GAF)What the GAF Tries to Capture
One of the most persistent criticisms of the GAF is that it asks a single number to do two separate jobs: rate the severity of a person’s psychiatric symptoms and rate how well they function in daily life. These are related but genuinely distinct things. Someone can have significant hallucinations yet still hold a job and maintain friendships, and someone else can have mild anxiety but be unable to leave the house. Forcing both dimensions into one score means something has to give.
Research confirms that when clinicians assign a GAF score, symptom severity tends to win. A study examining how Axis V was used in practice found that symptom ratings were generally worse than functional impairment ratings, and the final GAF score tracked more closely with symptom severity than with how well a person was actually managing their life. Specifically, the symptom component correlated more strongly with positive symptoms like hallucinations and delusions, while functional ratings were more connected to negative symptoms and real-world impairment.
2PubMed. The assessment of symptom severity and functional impairment with DSM-IV axis VThis is not just a theoretical quibble. If a patient’s GAF score is driven mostly by symptoms rather than actual functioning, treatment decisions, disability determinations, and discharge planning may all be shaped by an incomplete picture. A person whose symptoms improve on medication might see their GAF climb even though they still cannot work or live independently, and that inflated score could affect whether they qualify for services.
The Reliability Problem
For a clinical measurement to be useful, different clinicians rating the same patient should arrive at roughly the same number. The GAF’s track record on this front is mixed at best.
When trained researchers rate patients using the GAF under controlled conditions, agreement can be quite good. One study of psychiatric staff responsible for assessing first-time outpatients found satisfactory inter-rater reliability, with an intraclass correlation coefficient of 0.81. The researchers noted that a clinician’s own attitude toward the GAF and their general motivation to use measurement tools in psychiatry were associated with how reliably they scored it.
3PubMed. Reliability of global assessment of functioning ratings made by clinical psychiatric staffThe picture looks different in everyday clinical practice. A study comparing routine clinical GAF scores with research-grade scores found that inter-rater reliability between clinicians and researchers was low, with intraclass correlation coefficients ranging from 0.39 to 0.59. Meanwhile, reliability between the two researchers themselves was excellent, at 0.81 to 0.85. The takeaway is that GAF scores are reproducible when the people using them are carefully trained, calibrated, and motivated, but those conditions often do not hold in a busy clinic.
4PubMed. Are GAF scores reliable in routine clinical use?A large study of outpatients with major depressive disorder showed a similar pattern. Clinicians assigned a mean GAF score that was systematically lower than the score a test nurse gave the same patients, and the correlation between the two was weak. The GAF showed poor inter-rater reliability and poor ability to distinguish between disease severity and physical limitations in that sample.
5PubMed. Reliability and validity of the Global Assessment of Functioning Scale in clinical outpatients with depressive disordersPart of the explanation is that the GAF absorbs the biases and tendencies of whoever is scoring it. A multilevel analysis of inpatient GAF scores found that about 7 to 8 percent of the residual variance in scores was attributable to the individual physician doing the rating, even after accounting for patient-level factors. That physician-level effect was statistically significant, meaning who rates you can shift your score in a way that has nothing to do with your actual condition.
6PubMed Central. Multilevel analysis of the determinants of the global assessment of functioning in an inpatient populationWhy It Was Dropped From the DSM-5
When the American Psychiatric Association published the DSM-5 in 2013, the entire multiaxial system was retired, and the GAF went with it. The decision was not made on a whim. The criticisms outlined above, the conflation of symptoms with functioning, unreliable scoring in routine practice, and the difficulty of capturing a complex clinical picture in one number, had accumulated over years of published research. The DSM-5 work groups concluded that a single global score was too blunt and too inconsistent to serve as a standard part of diagnosis.
The GAF also had a conceptual problem that went beyond measurement. Its scoring anchors mix psychiatric symptoms with social and occupational functioning in ways that make it hard for clinicians to know what they are actually rating. A score of 50 could reflect serious symptoms with moderate functioning, moderate symptoms with serious functional impairment, or some blend of both. The ambiguity baked into the scale meant that two clinicians could agree on a number but disagree entirely on what it meant for the patient’s life.
7BioMed Central / Annals of General Psychiatry. Guidelines for rating Global Assessment of Functioning (GAF)What Replaced It
The DSM-5 recommended the World Health Organization Disability Assessment Schedule 2.0, or WHODAS 2.0, as the successor for measuring disability and functioning. Unlike the GAF, the WHODAS 2.0 is not psychiatry-specific. It was designed by the WHO to assess disability across all health conditions, with domains covering cognition, mobility, self-care, getting along with others, life activities, and participation in society. It also comes in a self-report version, meaning patients can complete it themselves rather than relying entirely on a clinician’s judgment.
Whether the WHODAS 2.0 actually measures the same thing the GAF measured is a separate question. A study comparing the two instruments in adults with schizophrenia spectrum disorders found no significant correlation between raw WHODAS 2.0 scores and GAF scores across all participants. When a clinician corrected the WHODAS 2.0 responses for outpatients specifically, a modest correlation emerged, but the self-rated version of the WHODAS 2.0 did not correlate with the GAF at all.
8PubMed. Comparing Global Assessment of Functioning (GAF) and World Health Organization Disability Assessment Schedule (WHODAS) 2.0 in schizophreniaThat disconnect is important because many systems, particularly the VA disability compensation process, relied on GAF score ranges to anchor benefit decisions. If the new tool does not map neatly onto the old one, transitioning from GAF-based criteria to WHODAS-based criteria is not straightforward. A study of veterans seeking compensation for PTSD concluded that the WHODAS 2.0 is an acceptable replacement for the GAF in that population, though “acceptable” reflects adequacy for the purpose of assessing functional impairment rather than perfect equivalence.
9PubMed. Using the WHODAS 2.0 to Assess Functioning Among Veterans Seeking Compensation for Posttraumatic Stress DisorderWhere the GAF Still Matters
Despite its removal from the DSM-5, the GAF has not vanished. It remains embedded in clinical systems and regulatory frameworks that were built around it. Many countries outside the United States continue to use it as a standard measure. Norway, for instance, requires GAF reporting in its mental health services. In the U.S. VA system, GAF scores assigned before the DSM-5 transition still appear in veterans’ medical records and can influence ongoing disability rating decisions. Courts and forensic evaluators sometimes still reference GAF scores, particularly for cases that originated when the scale was standard practice.
In forensic psychiatry, the GAF has been used alongside newer instruments to assess functioning in patients detained in secure settings. A validation study of a capacity assessment tool for forensic patients with schizophrenia and schizoaffective disorder found significant and large correlations between the new tool and the GAF across occupational, social, and symptomatic domains. Patients at lower levels of therapeutic security, meaning those closer to community release, showed greater functional capacity on both measures.
10PubMed Central. An evaluation of functional mental capacity in forensic mental health practice: the Dundrum capacity ladders validation studyThe GAF also remains a common outcome measure in research on inpatient psychiatric care. A study examining the effectiveness of psychiatric hospitalization found a mean GAF improvement of about 21 points over the course of an inpatient stay, with a large overall effect size. The improvement varied by diagnosis: substance-related disorders showed the smallest gains, while other mood disorders showed the largest. Length of stay and which ward a patient was on also influenced the degree of change.
11PubMed Central. Effectiveness of psychiatric inpatient careThe Children’s Version
The GAF was designed for adults, but a parallel instrument called the Children’s Global Assessment Scale, or CGAS, was adapted from it for use with children and adolescents. Like the GAF, it runs from 1 to 100, with clinicians evaluating a child’s overall level of functioning across ten sections. Higher scores indicate better functioning.
The CGAS has shown more favorable psychometric properties than its adult counterpart. Early validation work found it to be reliable between raters and stable across time, and it demonstrated both the ability to distinguish between different diagnostic groups and agreement with other measures of severity. The developers recommended it as a useful complement to disorder-specific scales for both clinicians and researchers.
12JAMA Psychiatry. A Children’s Global Assessment Scale (CGAS)Studies of children and adolescents in psychiatric inpatient settings have used the CGAS as a primary measure of whether treatment is working. Its role there mirrors the GAF’s role in adult inpatient research: a single number that can be tracked over time to quantify improvement or decline.
13PubMed Central. Factors affecting improvement of children and adolescents who were treated in the child and adolescent psychiatry inpatient unitOther Scales in the Mix
The WHODAS 2.0 is not the only alternative to the GAF. In the United Kingdom and several other countries, the Health of the Nation Outcome Scales, or HoNOS, has been widely adopted as a routine outcome measure in mental health services. Unlike the GAF’s single number, HoNOS uses 12 separate items covering areas like overactive behavior, substance use, cognitive problems, physical illness, and social functioning. Each item is scored from 0 to 4.
How well HoNOS and GAF agree depends on the study you look at. One implementation study in an outpatient clinic found a strong negative correlation between the two, which is the expected direction since higher HoNOS scores mean worse outcomes while higher GAF scores mean better functioning.
14European Psychiatry. Implementation of Health of the Nation Outcome Scale (HoNOS) in Outpatient Clinic, Sligo Mental Health Services: Feasibility and Agreement with Global Assessment of Functioning Scale A later study from the same service reported even stronger agreement between the two instruments.
15PubMed. Implementation of outcome measurement (HoNOS) in an outpatient psychiatric clinic in Sligo/Leitrim mental health serviceThe advantage of multi-item scales like HoNOS or the WHODAS 2.0 is that they do not force a complex clinical picture into a single number. A patient might score well on self-care but poorly on social relationships, and that profile is preserved rather than averaged away. The disadvantage is convenience. A single number is easy to record, compare, and communicate. It is also easy to misuse, but its simplicity is exactly why the GAF persisted for as long as it did and why clinicians sometimes still reach for it informally even when it is no longer required.
How a GAF Score Can Affect Your Life
If you are a veteran filing a disability claim, a GAF score in your records can carry real weight. The VA historically used GAF ranges to help determine disability percentages, and while the system has been transitioning to other measures, older scores still factor into appeals and rating decisions. A GAF of 50, for instance, puts you in the “serious symptoms” range and might support a higher disability rating than a GAF of 65, which falls in the “some mild symptoms” territory. Because the difference between these numbers can translate directly into monthly compensation, the unreliability of the score becomes a practical problem, not just an academic one.
In court settings, a GAF score can appear in competency evaluations, sentencing recommendations, or civil commitment hearings. A low score can support an argument that someone needs a higher level of care or is not fit to stand trial. A high score can be used to argue the opposite. The problem, again, is that two qualified clinicians might assign meaningfully different scores to the same person, and neither would necessarily be wrong given the scale’s vagueness.
For patients in routine psychiatric care, the GAF score usually matters less directly. Your treatment plan is shaped by your specific symptoms, history, and goals, not by a single number. But GAF scores can influence referrals, determine eligibility for certain programs, and shape how your case is described in records that follow you between providers. Understanding that the number is an approximation, and that it leans more toward your symptoms than your actual daily functioning, is worth knowing if you ever see it in your own chart.
What Clinicians Wish Patients Knew
The GAF score was never meant to be a comprehensive description of a person’s mental health. It was designed as a quick snapshot, a way to communicate roughly how someone is doing at a particular moment. The fact that it became embedded in high-stakes decisions about disability benefits, legal competency, and insurance coverage says more about the systems that adopted it than about the scale itself.
If you encounter a GAF score in your medical records or a disability evaluation, treat it as one data point rather than a definitive judgment. A low score does not mean you cannot improve, and a high score does not mean you are fine. The score reflects one clinician’s impression at one moment in time, shaped by the information available and influenced to some degree by the clinician’s own tendencies. Asking your provider what went into the rating and whether it captures your experience accurately is a reasonable thing to do, and most clinicians will appreciate the question.