Sign Up
Login
Try for Free
Login
The GAF (Global Assessment of Functioning) is a clinician-rated 1 to 100 score of psychiatric symptom severity and functioning, published as Axis V of DSM-IV-TR and dropped from DSM-5 in 2013. Psychiatrists, psychologists, and community mental health clinicians still enter one when a payer form, legacy EHR field, disability file, or court asks for it. This page covers how to document a defensible legacy GAF, with a fictional sample note.
Psychiatrists, psychologists, psychiatric nurse practitioners, and licensed mental health clinicians who assign the rating from the interview and the record; no publisher qualification level, license, or training requirement applies, and the APA offers no current training or license for it
Payers and utilization reviewers with legacy forms, state Medicaid and community mental health systems, SSA and VA adjudicators reading historical records, courts and attorneys, disability and workers' compensation evaluators, later treating clinicians
2 to 6 chart lines (edition frame and reason, period rated, basis, supporting facts, companion measure) · rating takes 1 to 3 minutes after the interview
Clinician-rated single global rating (1 to 100 in ten conceptual bands, 0 for inadequate information; symptom severity and functioning rated together, with the lower of the two governing)
Legacy Axis V fields in payer authorization and utilization review forms, state Medicaid and community mental health level-of-care records, disability, workers' compensation, and forensic reports, historical record review, and EHR templates never migrated to DSM-5
DSM-IV-TR Axis V (American Psychiatric Association, 2000), copyrighted; removed from DSM-5 in 2013 with the WHODAS 2.0 placed in Section III; no federal authority requires it, some state rules and payer forms still do; described here for documentation, no anchor text reproduced
The Global Assessment of Functioning is a single clinician-rated number, 1 to 100, that summarizes a person's psychiatric symptom severity and psychological, social, and occupational functioning, with 0 reserved for inadequate information. It descends from the Global Assessment Scale of Endicott, Spitzer, Fleiss, and Cohen (1976), entered the DSM as Axis V in DSM-III-R (1987), and was carried through DSM-IV (1994) and DSM-IV-TR (2000), which is the last edition to contain it. The scale is divided into ten conceptual ten-point bands, each pairing a level of symptom severity with a parallel level of functioning, running from a symptom-free, high-functioning top of the scale to bottom bands where safety or basic self-care cannot be maintained; the exact wording of each band is the American Psychiatric Association's copyrighted text and is not reproduced here. Three rating instructions matter for documentation: the rating describes the current period unless another is specified, and clinicians have operationalized current as today, the past week, or the past month; when symptom severity and functioning fall in different bands, the lower of the two governs rather than an average; and impairment caused by physical or environmental limitations is excluded from the rating. DSM-5 (2013) removed the multiaxial system and the GAF, citing its lack of conceptual clarity (one number mixing symptoms, suicide risk, and disability) and questionable psychometrics in routine practice, and placed the WHO Disability Assessment Schedule 2.0 in Section III as an emerging measure rather than a mandated replacement; DSM-5-TR (2022) kept that arrangement.
The load-bearing fact for documentation is that the GAF is retired from diagnosis but not from administration. No clinical body requires it in 2026, the VA stopped using it for disability examinations in 2014, and the Social Security Administration treats it as one piece of opinion evidence, yet a Rhode Island Medicaid rehabilitative-services guideline updated in April 2026 still asks for a GAF on Axis V, Texas level-of-care guidelines still carry a GAF intake marker beside the ANSA, a Washington regulation effective July 2025 still hard-codes a DSM-IV GAF screening rule for one counseling credential, Wisconsin still hosts a 2010 prior-authorization form with a current-GAF field, and payer utilization-review forms, legacy EHR templates, disability files, and court records keep asking for a number. That gap is what a defensible entry has to bridge: the note must say which edition's conventions produced the number, what period it covers, whether symptoms or functioning drove it, what facts support it, and what current functional measure sits beside it, usually the WHODAS 2.0, which measures disability across six domains and does not convert to a GAF. The modified GAF (Hall, 1995), the split GAF-S and GAF-F (Karterud and colleagues, 1998), the Kennedy Axis V (2003), the VA's MIRECC GAF (Niv and colleagues, 2007), and the Children's Global Assessment Scale (Shaffer and colleagues, 1983) are separate instruments with their own provenance, and none is a DSM-IV-TR GAF. The diagnostic structure around the entry belongs to the psychiatric diagnostic evaluation; the goals it is sometimes asked to justify belong to the treatment plan.
Community mental health and public-sector clinicians meet the GAF most often: psychiatrists, psychiatric nurse practitioners, psychologists, and licensed clinical social workers completing state Medicaid or managed-care authorization forms that never lost their Axis V field, level-of-care records that still carry a GAF marker, and EHR intake templates with a mandatory numeric field. Utilization reviewers and prior-authorization staff read those entries, so the number and its explanation travel into the utilization review summary and the prior authorization request. Disability, workers' compensation, and forensic evaluators supply one when a referral form, a state schedule, or a tribunal asks, and then explain its limits in the medical source statement or the court report, where the legally useful content is the concrete functional limitation, not the number. VA and SSA adjudicators encounter GAFs in older records and weigh them as history. In every one of those settings the GAF loses to its neighbors as the functional measure: the WHODAS 2.0 for a current, domain-level disability score, the mental status exam for the observed symptom picture the number is supposed to summarize, and the system's own instrument (ANSA, CANS, LOCUS, HoNOS) where one exists. The GAF page owns one narrow job: how to write the retired number so that it cannot be misread.
No authority publishes a 2026 GAF documentation standard, because the scale is no longer part of any current diagnostic manual; what exists is the DSM-IV-TR rating instructions, the SSA's 2013 caution that evaluators rarely say whether a score reflects symptoms or functioning, and the forms that still demand a number. What survives review is an entry that names the edition frame and the reason for the entry, states the period rated, discloses which component drove the number, gives the supporting facts in a sentence, enters one integer with no threshold claim, records a current functional measure beside it, and treats prior scores as dated opinions. Each element below carries the pitfall that most often undermines it.
Edition frame and reason for the entry. Open with the convention and the trigger: "GAF rated per DSM-IV-TR conventions (APA, 2000); scale removed from DSM-5 in 2013; entered because the payer's continued-stay form requires an Axis V value." That one sentence tells the next reader that the number is a legacy rating supplied on request, not a current DSM-5-TR measure, and it names the authority that asked, which is the first thing a reviewer or attorney will want to know. If the field is a software validation rule rather than a policy requirement, say so. Pitfall: A bare "GAF: 55" in a 2026 note, read by the next clinician, the payer, or the court as a current standardized measure that the DSM still endorses.
Period rated. State the window the number describes: at today's evaluation, the past week, the past month, or a defined date range. If the form or your practice also records the highest level sustained in the past year, label it separately; DSM-IV practice treated current and highest-past-year as different questions, and SSA's guidance notes that the higher sustained level is the better prognostic signal. A retrospective number for a period you did not examine is an inference, and the note should say so or omit it. Pitfall: Current and past-year values mixed in one field, or a present-tense number assigned to a period the rater never observed.
Basis: symptoms, functioning, or both. Under DSM-IV-TR conventions the overall number reflects the worse of the two components, never their average. Write which side governed: "symptom severity is the limiting component; occupational and self-care functioning are comparatively preserved," or the reverse, or "symptoms and functioning fall in the same band." This is the disclosure the SSA found evaluators rarely make, and it is what stops a symptom-driven number from being read as global incapacity. If a split rating (GAF-S and GAF-F) or a modified version was used, name it; a split score is not a DSM-IV-TR GAF and should not be entered as one. Pitfall: Serious symptoms and an intact job averaged into the middle of the scale, or a number with no indication of which dimension produced it.
Supporting facts, in one sentence. Give the observations a reviewer can weigh: work or school attendance, independent living and self-care, relationships and withdrawal, psychotic symptoms, cognitive organization, and suicidal ideation with a pointer to the documented risk assessment. Keep limitations from physical illness, pain, transportation, housing, or the labor market out of the rating and note them elsewhere; the GAF excludes them by instruction, and a low psychiatric number built from an orthopedic or housing problem is the error disability reviewers look for. Pitfall: A number with no facts under it, or a physical or environmental limitation silently pulling the psychiatric rating down.
The number itself. Enter one integer, not a range, and describe the level in your own words rather than copying anchor text into the chart. Do not attach a level-of-care or medical-necessity conclusion to the number unless a named policy actually sets that threshold, and then cite the policy; unofficial guides that map GAF values to inpatient, intensive outpatient, or outpatient care tie to no authority. Remember the reliability record: routine-practice raters agree with trained raters only moderately, so the number is an estimate with error around it, and the note should read that way. Pitfall: "GAF 45 to 50," or "GAF 48, meets criteria for intensive outpatient" with no policy behind the threshold.
Companion current measure. Record a present-day functional measure beside the GAF with its own date and result: the WHODAS 2.0 (36- or 12-item, self, proxy, or interviewer version, 30-day recall) for a DSM-oriented adult evaluation, or the instrument the system already runs (ANSA, CANS, LOCUS, HoNOS, the K10 in Australian services). Keep the two analytically separate: there is no validated conversion between a WHODAS total and a GAF value, and the note should say the GAF is supplied for the form while the companion measure carries the functional evidence. Pitfall: A WHODAS domain profile translated into a GAF number, or the two scales charted as one trend line.
Historical scores and change. Carry prior GAFs forward as what they are: a dated opinion by a named rater about a stated period. Quote them with the date and author, do not re-date them as current, and do not present the difference between two raters' numbers as treatment response; the SSA's own guidance warns that differences between clinicians' GAFs cannot be read as reliable change, and the published inter-rater figures explain why. Describe change in the facts (attendance, symptoms, risk, self-care) and let the number follow. Pitfall: "Improved from 42 to 55 since intake" across two raters, or a 2011 GAF re-entered in a 2026 field without its date.
GAF DOCUMENTATION BLOCK (legacy Axis V entry) Date: [ ] Setting: [ ] Clinician: [ ] Encounter: [intake / continued-stay authorization / disability or forensic report / review] Edition frame: [rated per DSM-IV-TR conventions (APA, 2000); scale removed from DSM-5 (2013)] Reason for entry: [form / field / order that requires it; policy or software validation rule] Period rated: [at today's evaluation / past week / past month / dates] Highest sustained in past year (if requested, separate): [ ] GAF value: [one integer, 1 to 100; 0 only for inadequate information] Version if not standard: [split GAF-S / GAF-F; modified; MIRECC; CGAS] Basis: [symptom severity / functioning / both in the same band is the limiting component] Supporting facts (one sentence): [work or school; self-care and living situation; relationships; psychotic symptoms; cognition; suicidal ideation with risk assessment reference] Excluded from the rating: [physical, pain, transportation, housing, labor-market limits, noted at ...] Companion current measure: [WHODAS 2.0 version and date, or ANSA / CANS / LOCUS / HoNOS / K10; result; no conversion to GAF] Historical GAFs: [value, date, rater, period rated; kept as dated opinions] Limits statement: [not a current DSM-5-TR measure; not normed; weigh with the narrative and the companion measure] Plan: [next review; what the entry supports] Clinician signature / credentials: Date:
Free to use and share, no signup. The PDF includes a one-page cheat sheet with element-by-element pitfalls and a pre-sign checklist; the DOCX is the blank documentation block, ready to adapt. Neither reproduces the scale's anchor descriptions.
Scenario: a community mental health center continued-stay authorization for an adult with recurrent major depressive disorder, on a state Medicaid form that still carries an Axis V field; the psychiatric nurse practitioner enters a symptom-driven legacy GAF, explains the basis, records a WHODAS 2.0 beside it, and carries the intake score forward as a dated opinion. All details are fictional.
Patient: A.K., 41 · Setting: Community mental health center, continued-stay authorization · Clinician: D. Reyes, PMHNP-BC · Note date: 09/10/2026
Frame and reason for entry: GAF rated per DSM-IV-TR conventions (APA, 2000); the scale was removed from DSM-5 in 2013 and is entered here only because the state Medicaid continued-stay authorization form requires an Axis V value. The rating is not a current DSM-5-TR measure and is not normed; the functional evidence for this request is the WHODAS 2.0 and the narrative below.
Rating: Current GAF, rated for functioning at today's evaluation on 09/10/2026: 45. No past-year value is requested by the form and none is estimated. Standard single rating; no split or modified version used.
Basis and supporting facts: Symptom severity is the limiting component. A.K. reports persistent depressed mood, early waking, and recurrent passive thoughts of death without plan or intent (risk assessment documented separately today, low acute risk, safety plan reviewed), with marked withdrawal from friends and family over the past month. Functioning is comparatively preserved: she has kept her part-time retail job with two late arrivals this month, lives alone, manages meals and medication, and attends appointments. Under the lower-of-the-two convention the symptom picture, not occupational capacity, produced the number. Chronic knee pain that limits standing is noted in the medical history and is excluded from this rating.
Companion current measure: WHODAS 2.0, 12-item, self-administered, 30-day recall, completed in the waiting room on 09/10/2026 and entered in the outcome-measure record: greatest difficulty reported in participation and life activities, moderate difficulty getting along with people, little difficulty with cognition, mobility, or self-care, consistent with the interview. No numerical conversion between the WHODAS and the GAF was made.
Historical scores: Intake on 08/18/2026, rated by the intake clinician (R. Nakamura, LCSW) for the preceding week: GAF 42. The two ratings come from different raters and are recorded as dated opinions, not as measured change; the clinical change since intake is described in the facts above (sleep and attendance slightly better, withdrawal and passive death thoughts unchanged).
Plan and record: Continued weekly psychotherapy and medication management requested for 90 days; treatment plan goals and measures are in the treatment plan dated 08/18/2026. Repeat the WHODAS 2.0 at the December review; the GAF will be re-entered only if the form still requires it, with the same frame. Entry stored in the authorization section of the record with the WHODAS in the outcome-measure record; A.K. was told the number is a legacy rating supplied for the payer's form.
This sample is fictional and for educational purposes. It does not describe a real patient or record; the scores, dates, and details are invented to show documentation structure and are not clinical guidance. No scale anchor text is reproduced.
Writing these after every session? BastionGPT drafts complete notes from bullets, dictation, or a transcript.
Generate a note from bulletsUnited States: the authority map is specific, and it points in two directions. Under LAW, the Department of Veterans Affairs adopted DSM-5 for its rating schedule by an interim final rule effective August 4, 2014 (79 FR 45093), stating that the APA had found the GAF of "limited usefulness in the assessment of the level of disability" and that examiners would stop reporting it without any change to the rating criteria, and the Court of Appeals for Veterans Claims held in Golden v. Shulkin, 29 Vet. App. 221 (February 23, 2018), that "the Board errs when it uses GAF scores to assign a psychiatric rating" where DSM-5 applies, while accepting that older scores remain minimally relevant as history. Under ADJUDICATIVE POLICY, the Social Security Administration's Administrative Message AM-13066 (effective July 22, 2013) treats a GAF as opinion evidence that needs supporting detail and a stated period, calls it a snapshot, warns that differences between clinicians' scores cannot be read as reliable change, notes that evaluators rarely say whether a score reflects symptoms or functioning, and states that "a GAF score is never dispositive of impairment severity"; for claims filed on or after March 27, 2017, medical opinions are weighed for supportability and consistency with no controlling weight, and adult mental impairments are rated on SSA's four functional areas, so no GAF value equals a listing or a residual functional capacity. At the state level the picture reverses. Rhode Island's Medicaid Rehabilitative Services Coverage Guidelines (page updated April 28, 2026) still require a functional narrative that supports "the GAF score entered on Axis V" for authorization (PAYER POLICY); Texas Resilience and Recovery adult utilization-management guidelines (effective April 2017, still posted in September 2026) carry a GAF at or below 50 at intake as a marker of significant functional impairment inside level-of-care criteria that otherwise run on the ANSA (PAYER POLICY, legacy marker); Washington's WAC 246-810-021 (effective July 1, 2025) requires the state's limited-scope adviser credential to screen functional impairment with the DSM-IV GAF and to refer any client scoring 60 or less to a licensed practitioner (LAW, for that credential only); and Wisconsin ForwardHealth still hosts prior-authorization form F-00212 (2010) with a current-GAF field while the form itself says its use is voluntary and optional (CONVENTION: the field exists, the requirement does not). Nationally, CMS's 2026 Behavioral Health Core Set contains no GAF measure, and the circulating charts that map GAF ranges to inpatient, intensive outpatient, or "medical necessity not met" tie to no named payer or regulation (CONVENTION, habit rather than policy). In civil, family, and workers' compensation litigation the number appears because the underlying records are old or the form is, and its weight turns on foundation, not on a cutoff (CONVENTION); California's pre-2013 permanent disability rating schedule, which converted a GAF to a whole-person impairment, is the kind of legacy schedule an evaluator should cite by name if it governs the case.
Canada and Australia have moved on, and the reliability record explains why the edition line matters everywhere. No Canadian federal or provincial law or payer names the GAF; Ontario's Workplace Safety and Insurance Board renamed its psychology program the mental health program of care on April 27, 2026 and asks for history, testing, functional status, a current DSM diagnosis, treatment planning, and a separate return-to-work form describing abilities and restrictions, with no GAF field (PAYER POLICY). In Australia the national outcomes and casemix collection is built on the HoNOS family with consumer measures such as the K10+ and the LSP-16, and the CGAS for children; the adult GAF is not part of that set (CONVENTION), NDIS access evidence is organized around functional capacity in everyday activities rather than an Axis V number (LAW), and no MBS item names a GAF (PAYER POLICY). On the evidence: Vatnaland and colleagues (2007) found routine clinical GAF scores agreed with research ratings only moderately (intraclass correlations of 0.39 to 0.59) while two trained researchers agreed at 0.81 and 0.85; Soderberg and colleagues (2005) reached an intraclass correlation of .81 when 81 staff rated first-time outpatient vignettes; Abbo and colleagues (2013) raised agreement on current ratings from 0.59 to 0.83 with one hour of training; Hall's modified GAF (1995) reported admission agreement of 0.81 against 0.62 for the original scale. Moos, McCoy, and Moos (2000) found in 1,688 patients that diagnoses and symptoms predicted GAF ratings more strongly than social and occupational functioning and that the ratings were only minimally associated with one-year outcomes; Pedersen and Karterud (2012) found substantial symptom-versus-function disagreement in about 10 percent of 2,695 patients, with functioning usually the worse of the two; and Suzuki and colleagues (2015) found a modest cross-sectional relationship with the PANSS across 40 schizophrenia trials (r of about -0.40, roughly 8,000 participants). Those numbers are the reason to write "per DSM-IV-TR conventions" with the date, disclose the basis, and treat any single value as an estimate with error around it; DSM-IV-TR (2000) is the only edition whose conventions a rater can cite, DSM-5 (2013) and DSM-5-TR (2022) contain no GAF, and the split, modified, MIRECC, Kennedy, and children's versions are separate instruments that must be named if used.
The GAF is copyrighted by the American Psychiatric Association as part of DSM-IV-TR (2000), and it is not made available under any general license: the scale was dropped from DSM-5 in 2013, the APA declined a documented request to reproduce it in March 2013 (recorded in the CamCOPS task documentation, which for that reason collects the number without the scale), and as of September 2026 no APA grant covering forms, EHR builds, or public web tools has been published, so the anchor descriptions should be treated as unlicensed text and paraphrased in original words wherever they are needed; the concept of a 1 to 100 rating is not protected, the APA's wording of the bands is. The WHODAS 2.0 is copyrighted by the World Health Organization, which licenses its inclusion in electronic records and any reproduction separately. BastionGPT is not affiliated with, or endorsed by, any of these publishers. This page reproduces no test items, stimuli, norms, or scoring materials.
The numbers behind these errors are specific. Routine clinical GAF scores agreed with trained research ratings at intraclass correlations of only 0.39 to 0.59, against 0.81 to 0.85 between two trained raters (Vatnaland and colleagues, 2007); one hour of training moved agreement on current ratings from 0.59 to 0.83 (Abbo and colleagues, 2013); in 1,688 patients, diagnoses and symptoms predicted the rating more strongly than social and occupational functioning, and the rating barely predicted one-year outcomes (Moos, McCoy, and Moos, 2000); symptom and function ratings diverged substantially in about 10 percent of 2,695 patients, usually with functioning worse (Pedersen and Karterud, 2012); and the Social Security Administration's 2013 guidance recorded that evaluators rarely note whether a score reflects symptoms, functioning, or both. No published audit counts how often a 2026 chart entry omits the edition frame or the period; the VA, SSA, and state policies described under compliance considerations are what turn those omissions into a weight problem. The BastionGPT Clinical Advisory Board sees the same errors most often in GAF documentation reviews:
BastionGPT is specifically trained, tuned, and clinically tested on psychiatric diagnostic evaluations and treatment plans.
See how clinicians use it day to day on the AI therapy notes page.
Many BastionGPT users report saving more than 90 minutes per day on documentation.
HIPAA-compliant with a signed BAA on every plan. Your data is never used to train models. BastionGPT drafts, you review and sign.
A GAF is a clinician's single judgment, on a 1 to 100 scale, of how severe a person's psychiatric symptoms are and how well they function psychologically, socially, and occupationally; 0 means there was not enough information to rate. Under DSM-IV-TR the scale is divided into ten-point bands, each pairing a description of symptom severity with a parallel description of functioning, stepping from a symptom-free, high-functioning top of the scale down to bands where safety or basic self-care cannot be maintained. When the two descriptions point to different bands the lower one governs, the rating covers the current period unless another is stated, and impairment from physical or environmental causes is left out. The band wording is the American Psychiatric Association's copyrighted text, so use the manual itself when you rate and describe the level in your own words in the chart. The number was never normed, and routine-practice raters agree only moderately, so read any single value as an estimate with error around it.
Only when something specific asks for it. DSM-5 (2013) and DSM-5-TR (2022) do not contain the scale, no clinical body requires it, the VA stopped using it for disability examinations in August 2014, and Social Security treats it as one non-dispositive opinion. What still asks: some state Medicaid and community mental health forms (Rhode Island's rehabilitative-services guideline, updated April 2026, still calls for a GAF on Axis V; Texas adult level-of-care guidelines still carry a GAF intake marker), one Washington regulation for a limited-scope counseling credential (effective July 2025), legacy prior-authorization forms such as Wisconsin's 2010 F-00212 (which says its own use is optional), payer utilization-review workflows, EHR templates never migrated to DSM-5, and disability or court files built on older records. The first documentation step is to find out which of those you are dealing with: a policy requirement gets a properly framed legacy entry; a software validation rule gets the same entry plus a note that the field is a system artifact; and where text is allowed instead of a number, a statement that the GAF is not a current DSM-5-TR measure, with a pointer to the functional assessment, is the better answer.
Answer five questions in one short paragraph. Which conventions: "rated per DSM-IV-TR conventions (APA, 2000); scale removed from DSM-5 in 2013; entered because [form or field] requires it." What period: today's evaluation, the past week, the past month, or a date range, with any highest-past-year value labeled separately. What drove the number: symptom severity, functioning, or both in the same band, never an average. What supports it: one sentence of facts (work, self-care, relationships, psychosis, cognition, suicidal ideation with the risk assessment referenced), with physical and environmental limits excluded. What current measure sits beside it: a dated WHODAS 2.0 or the system's own instrument, with an explicit statement that no conversion was made. Close with a limits line: not a current DSM-5-TR measure, not normed, weigh with the narrative. The sample on this page shows the five parts in a continued-stay authorization; the blank template carries them as fields.
Under DSM-IV-TR conventions the symptoms do: when symptom severity and functioning fall in different bands, the overall rating reflects the lower of the two, so a patient with active suicidal ideation or psychotic symptoms and an intact job receives a symptom-driven number, and a patient with modest symptoms and a collapsed work and social life receives a functioning-driven one. Neither is averaged. The documentation fix is to say so: "symptom severity is the limiting component; occupational and self-care functioning are comparatively preserved," followed by the facts, so a reviewer does not read the low number as global incapacity or the high one as absence of illness. If your setting uses the split GAF (a separate symptom score and functioning score), record both under their own labels and do not enter either in a field that asks for a standard DSM-IV-TR GAF without saying which it is. Pedersen and Karterud found the two dimensions diverged substantially in about one patient in ten, and functioning was usually the worse of the two, which is why the disclosure is worth a sentence every time.
Report it as history, never as a current opinion, and do not compute change from it. Quote the prior score with its date, its rater, and the period it described, keep it in a historical-scores line, and let your own current rating stand on its own frame. Two numbers from two raters are not a measurement of change: routine clinical raters agreed with trained raters at intraclass correlations of only 0.39 to 0.59 in Vatnaland and colleagues' 2007 study, and SSA's 2013 guidance tells adjudicators not to assume that differences between clinicians' scores reflect real change. Even two ratings by the same clinician describe two snapshots of possibly different periods. VA case law treats older GAFs as minimally relevant psychiatric history that cannot drive a DSM-5-era rating, and SSA treats them as opinion evidence that needs supporting detail. Describe change in the facts (attendance, self-care, symptoms, risk) and, if the form insists on a comparison, say the two values come from different raters or periods and are not a trend.
No and no. The GAF is a clinician's single global judgment that blends symptom severity with functioning and lets the worse component dominate. The WHODAS 2.0 is the World Health Organization's disability measure, built on the International Classification of Functioning, that asks about difficulty over the past 30 days across six domains (cognition, mobility, self-care, getting along, life activities, participation) in 36- or 12-item versions, self, proxy, or interviewer administered; DSM-5 placed it in Section III as an emerging measure, not as a mandated replacement. The two answer different questions, no validated crosswalk between a WHODAS total and a GAF value exists, and a note that converts one into the other invents data. The defensible pattern is side by side: the WHODAS carries the current, domain-level functional evidence; the GAF, when a form demands it, is entered with its frame, period, basis, and limits; and the entry says no conversion was made. In systems that run their own instrument (ANSA, CANS, LOCUS, HoNOS) the same rule applies with that instrument in the WHODAS's place.
Not to decide anything. The VA replaced DSM-IV references in its rating schedule with DSM-5 by an interim final rule effective August 4, 2014 (79 FR 45093), stating that examiners would no longer report the GAF and that the change affected only the form of their findings, not the rating criteria; in Golden v. Shulkin (2018) the Court of Appeals for Veterans Claims held that the Board errs when it uses GAF scores to assign a psychiatric rating in cases where DSM-5 applies, while accepting that older scores remain minimally relevant as history. Social Security's AM-13066 (2013) treats a GAF as opinion evidence that needs supporting detail and a stated period, calls it a snapshot, and says it is never dispositive of impairment severity; for claims filed on or after March 27, 2017, medical opinions are weighed for supportability and consistency with no controlling weight, and adult mental impairments are rated on SSA's four functional areas (understanding, remembering, and applying information; interacting with others; concentrating, persisting, and maintaining pace; adapting and managing oneself). So no GAF value equals a VA percentage, an SSA listing, or a residual functional capacity. For a disability report, document the concrete limitations in those terms in the medical source statement, and if a form still demands a GAF, enter it with its frame and limits rather than as the opinion itself.
Not the anchor descriptions, without written permission. The GAF is copyrighted by the American Psychiatric Association as part of DSM-IV-TR; the APA declined a documented request to reproduce the scale in March 2013 (the CamCOPS project records the refusal and collects the number without the scale for that reason), and no general APA grant covering forms, EHR builds, or web tools has been published as of September 2026. Being free, non-commercial, or attributed does not create a license, and the widespread claim that the GAF is public domain is not supported by anything the APA has published. What you may do is implement the concept: a numeric 1 to 100 field, your own independently written prompts about symptoms and functioning, and the documentation frame on this page; copyright protects the APA's wording of the bands, not the idea of a global rating. Two cautions: the WHODAS 2.0 is not a rights-free substitute (the WHO licenses its inclusion in electronic records and any reproduction), and a tool that stores a bare GAF number should also capture the period, basis, and frame, or it will generate exactly the undocumented numbers this page exists to prevent.
Yes. Give it the facts (the form or field that requires the number, the period rated, the symptom picture, the functional facts, the risk assessment reference, the companion measure and its result, and any prior scores with their dates and raters) and it drafts the full entry: the DSM-IV-TR frame and reason, the period, the basis with its supporting sentence, the value, the companion measure with the no-conversion statement, the historical scores as dated opinions, and the limits line, ready for your review. It can also check a finished note for a bare number with no frame or period, an undisclosed driver, a physical limitation inside the rating, a threshold claim with no policy behind it, or a WHODAS converted into a GAF. BastionGPT is HIPAA-compliant with a signed BAA on every plan, and your data is never used to train models.
The instrument facts and compliance claims on this page trace to these sources, last verified September 2026:
Educational content, not legal or billing advice. Sample notes are fictional. Follow your organization's policies and your board, payer, and jurisdiction requirements.