Sign Up
Login
Try for Free
Login
The MMSE (Mini-Mental State Examination) is a 30-point cognitive screen published by Folstein, Folstein, and McHugh in 1975, now a licensed instrument sold per form by its publisher. Hospitals, clinics, and drug-coverage programs still key decisions to it. This page covers how to document and interpret MMSE results in the chart, with a fictional sample note.
Credentialed healthcare professionals using purchased or licensed forms (publisher qualification level S)
Treating and referring clinicians, drug-coverage programs in Canada and Australia, courts and capacity evaluators, families
3 to 8 chart lines · administration 10 to 15 minutes
Clinician-administered cognitive screen (30 points; second-edition versions score 16, 30, and 90)
Cognitive screening and serial dementia tracking, drug-subsidy documentation in Canada and Australia, inpatient cognitive baselines
Licensed PAR instrument sold per form; Canadian and Australian drug plans still key coverage to its bands; no US authority mandates it
The Mini-Mental State Examination (MMSE) is a 30-point, clinician-administered cognitive screen published by Marshal Folstein, Susan Folstein, and Paul McHugh in 1975. The fact that shapes everything else about using it today is legal, not clinical: after circulating freely for about 25 years, its intellectual property was registered to MiniMental LLC in 2000, and in 2001 Psychological Assessment Resources (PAR) received the exclusive worldwide license. Enforcement followed, including the 2011 withdrawal of the free Sweet 16 screen after an infringement demand. Under the publisher's current terms, forms are purchased consumables (US list prices in August 2026: $2.50 per original-MMSE form, with second-edition forms priced per version), a purchased copy does not authorize photocopying, and reproducing prompts, stimuli, or scoring rules in an EHR template, an improvised translation, or administration "from memory" all fall outside the stated permissions. What is ordinary clinical documentation: recording the edition, version, score, date, limitations, and interpretation in the chart.
The scoring facts are just as easy to get wrong. The original and the second edition's Standard Version total 30 points, and the publisher equated them so longitudinal data carry across; the second edition's Brief Version totals 16 and its Expanded Version 90, so "MMSE-2 score" means nothing without the version, and each version ships in Blue and Red alternate forms. The famous cutoff of 24 is a convention built on a 23/24 boundary from early validation work, and it bends with demographics: in the largest US population study, the median score for adults with zero to four years of schooling was 22, which is why a low score in a low-education patient can be population-typical while a "normal" 27 can represent real decline in a highly educated one. The MMSE is also one letter away from the mental status exam, the descriptive clinical examination it is routinely confused with, and one comparison away from the MoCA, which detects mild impairment the MMSE misses. It is a screen; it diagnoses nothing alone.
Hospital and clinic teams document MMSE results as cognitive baselines and serial dementia measurements; the payer reality is where the score still carries hard weight. Canadian provincial drug plans key cholinesterase-inhibitor coverage to score bands (Ontario and British Columbia use 10 to 26, with BC on the Standardised MMSE plus a staging scale, while Alberta's 2026 criteria accept the MMSE or SLUMS, RUDAS, or interRAI alternatives). Australia's PBS requires a baseline of 10 or higher for subsidized cholinesterase inhibitors (10 to 14 for memantine) with specialist confirmation, runs a formal low-score pathway for patients whose scores reflect language, education, cultural, sensory, or dysphasia barriers rather than disease severity, and judges continuation on a documented clinically meaningful response, having abolished its old two-point-improvement rule in 2013. In the US, no Medicare rule names it: wellness-visit cognition detection and capacity and delirium workflows are all tool-neutral, and delirium pathways use delirium-specific instruments. When the question outgrows a screen, the next steps are an RBANS or a full neuropsychological evaluation.
No statute prescribes an MMSE note format. The elements below are what keeps a licensed, versioned, convention-laden screen interpretable and defensible over time. Each carries the pitfall that most often undermines it.
Edition, version, and form. Name the instrument precisely: original MMSE or MMSE-2, the version (Brief, Standard, Expanded), the Blue or Red form where applicable, and the authorized language edition. Pitfall: "MMSE 22." The three second-edition versions total 16, 30, and 90 points, only the Standard Version is score-equated to the original, and an unnamed version makes the number and every later comparison ambiguous.
The attention method, for legacy administrations. The original instrument permits two alternative attention tasks, and research shows the serial-subtraction and reverse-spelling approaches are not interchangeable: the arithmetic version runs harder and both are education-sensitive. Pitfall: a method switch between administrations charted as cognitive change. Record which method was used, every time.
Raw total over its denominator, with demographic context. Write the fraction and the interpretive context: completed education, literacy where material, language and interpreter use, and sensory or motor limits with aids in place. Pitfall: reading a low score in a low-education patient as disease. In the major US population norms, the median score with zero to four years of schooling was 22; education, language, and culture move totals without any pathology.
Validity and completeness. State whether administration was complete, cooperative, and standard; document refusals, interruptions, and untestable components with reasons. Pitfall: silent zeros and improvised prorating. No generally authorized rule converts an incomplete administration into a standard 30-point total; "administration incomplete; standard total not interpretable" is the defensible entry.
Domain pattern and function. Describe the broad pattern in your own words (orientation, registration, attention, delayed retrieval, language, construction) and tie it to function and collateral report. Pitfall: treating identical totals as identical patients. The total is a composite; very different cognitive patterns produce the same number, and function is what separates them.
Interpretation against the conventional threshold. Below 24 means below the conventional screening threshold, a flag for evaluation. Pitfall: verdict language. The 23/24 boundary comes from early validation work whose authors said the MMSE alone cannot diagnose, and ceiling effects mean highly educated patients with real impairment routinely score "normal": in one highly educated cohort, the standard cutoff caught barely more than half of dementia cases.
Serial comparison discipline. Compare only like with like: same edition, version, language, attention method, sensory aids, and clinical state, on alternate forms where available, with the interval stated. Pitfall: over-reading small movements. Reliable-change estimates run from roughly two to four points in some cohorts to five to seven in others, while average Alzheimer-type decline runs about three points a year, so a two-point drop is usually noise unless function moves with it.
MMSE DOCUMENTATION BLOCK Edition: [original MMSE / MMSE-2] Version: [Brief 16 / Standard 30 / Expanded 90] Form: [Blue / Red / n/a] Language (authorized edition): [ ] Interpreter: [ ] Attention method (legacy original): [serial subtraction / reverse spelling] Date: [ ] Setting: [ ] Administered by: [ ] Raw total: [ ]/[30 / 16 / 90] Context: [education years; literacy; sensory aids in place; motor limits; acute illness or delirium risk] Validity: [complete and standard / incomplete: which parts and why; refusals documented, not scored as errors] Domain pattern (own words): [orientation / registration / attention / delayed retrieval / language / construction] Function and collateral: [ADLs, IADLs, informant report] Interpretation: [screening result vs conventional threshold in demographic context; not a diagnosis] Prior: [date, edition/version/form, method, score] Change: [points; reliable-change and condition-comparability check] Plan: [workup / referral / reassessment interval] 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 any MMSE test content.
Scenario: an outpatient serial comparison, the use case where version, form, and method discipline earn their keep. This entry documents results and interpretation only; it reproduces no test content. All details are fictional.
Patient: W.K., 79 · Visit: Memory clinic follow-up · Clinician: P. Lindqvist, MD (geriatrics) · Note date: 08/11/2026
Measure: MMSE-2 Standard Version, Red form, English (authorized edition), administered in clinic by this clinician: 21/30. Education 14 years. Hearing aids and reading glasses in place; no acute illness; administration complete and standard. Purchased protocol used; this note records results only.
Prior: MMSE-2 Standard Version, Blue form, English, this clinic, 06/09/2025: 25/30, same conditions and aids. Interval 14 months; alternate forms used to limit practice effects; versions are score-equated, so the totals are comparable.
Pattern and function: Current performance was relatively weaker in orientation and delayed retrieval, with language and construction comparatively preserved. W.K.'s son reports new help with bills and two missed appointments since spring; medication management now supervised.
Interpretation and plan: The four-point decline may exceed short-term measurement variability in some cohorts, and the interpretation rests principally on its concordance with the reported functional decline rather than the number alone. The result is a screening measurement, not a diagnosis or a stage. Plan: laboratory review of reversible contributors, medication reconciliation, structured functional assessment, and discussion of fuller neuropsychological evaluation; reassess with the Blue form in 12 months under the same conditions, sooner on functional change.
This sample is fictional and for educational purposes. It does not describe a real patient or record, reproduces no MMSE test content, and the scores are invented for illustration and correspond to no real person.
Writing these after every session? BastionGPT drafts complete notes from bullets, dictation, or a transcript.
Generate a note from bulletsThe MMSE's compliance layer is unusual for a bedside screen: the operative questions are publisher licensing first and payer policy second. Under PAR's stated terms, authorized protocols are purchased consumables; a purchased copy does not authorize photocopying, downloading a "free" online copy, reproducing any part of the instrument in an EHR template, modifying, reformatting, or translating it, and the publisher's enforcement history is concrete (the free Sweet 16 screen was withdrawn in 2011 after an infringement demand). Academic commentary has long questioned how far copyright can reach into a clinical method, and no court judgment on the merits was located, but a dispute in the literature is not a permission slip: the low-risk operational course is purchased or licensed materials, with the distinction that matters for records kept clean. Documenting the edition, version, form, score, date, limitations, and interpretation is ordinary clinical charting; embedding prompts, stimuli, or scoring keys in a template is reproduction. In the US, the decision contexts are tool-neutral: the Medicare annual wellness visit requires detection of cognitive impairment by observation and report without naming an instrument, the cognitive assessment and care plan service (99483) requires a far larger documentation set than any score, delirium pathways call for delirium-specific instruments rather than an MMSE threshold, and capacity is decision-specific: the Department of Justice's capacity guide says no single test result should drive the determination, and the American Bar Association's guidance for lawyers puts it more bluntly: "Thou shalt not covet the mini-mental status exam." A score can inform a capacity evaluation; it can never be one.
Where the score still gates money, precision matters. In Canada, Ontario's drug benefit requires an MMSE of 10 to 26 to start and continue cholinesterase-inhibitor coverage; British Columbia keys donepezil to a Standardised MMSE of 10 to 26 plus Global Deterioration Scale stage 4 to 6 (and note that SMMSE means the Standardised MMSE, a different thing from the Severe MMSE used below the ordinary floor); Alberta's April 2026 criteria accept an MMSE of 10 to 26 or, notably, SLUMS, RUDAS, or interRAI Cognitive Performance Scale ranges instead, a payer formally accommodating instrument bias; Manitoba dropped the score gate for donepezil and galantamine back in 2018. Australia's PBS instrument requires a baseline MMSE or Standardised MMSE of at least 10 for cholinesterase inhibitors (10 to 14 for memantine) with specialist confirmation, runs an explicit low-score pathway when a score of 9 or less reflects language, education, Indigenous cultural factors, intellectual disability, sensory impairment, or disproportionate dysphasia rather than disease severity (with a clinician's-impression assessment instead), and since 2013 judges continuation on a documented, six-monthly, clinically meaningful response, having abolished the old two-point-improvement rule its own advisory committee found unreliable. Two final measurement cautions: a 2025 central review of 10,203 administrations flagged administration problems in 26.8% and scoring problems in 27%, so standardized administration is a live quality issue, not a formality; and published MoCA crosswalks are estimates for context, never substitutes for an administered score a payer requires.
MMSE is a registered trademark of Psychological Assessment Resources, Inc.; MMSE materials are copyright MiniMental LLC. BastionGPT is not affiliated with, or endorsed by, the publisher. This page reproduces no test items, stimuli, norms, or scoring materials.
The quality data are blunt: a 2025 central review of 10,203 MMSE administrations flagged administration problems in 26.8% and scoring problems in 27.0%. Interpretation has its own traps: in a highly educated cohort the standard cutoff caught barely more than half of dementia cases, while the major US population norms put the median score for adults with zero to four years of schooling at 22, inside the range routinely labeled impaired. The BastionGPT Clinical Advisory Board sees the same errors most often in MMSE documentation reviews:
BastionGPT is specifically trained, tuned, and clinically tested on behavioral health progress notes and screening documentation.
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.
It is a 30-point screening composite, and the familiar cutoff of 24 is a convention built on a 23/24 boundary from early validation work, where the authors themselves said the instrument cannot diagnose alone. Demographics move the number: in the largest US population norms, the median score was 29 with nine or more years of schooling and 22 with zero to four years, and scores drift down with age. A Cochrane review found reasonable but threshold-dependent accuracy in older community populations. So the defensible reading is "below the conventional screening threshold, in this demographic context, warranting evaluation," never a diagnosis or a stage.
No and no, under the publisher's stated terms. The instrument circulated freely for decades, but its rights were consolidated in 2000 and exclusively licensed to PAR in 2001, and enforcement is real (the free Sweet 16 screen was withdrawn in 2011 after an infringement demand). Authorized protocols are purchased consumables (US list price for the original ran $2.50 per form in August 2026), and the publisher's position is that purchase does not authorize photocopying, free-download copies, modification, or improvised translation. Legal commentators have questioned the copyright's reach, but no merits ruling settles it, so the low-risk course for a clinic is purchased or licensed materials, or a free alternative instrument chosen deliberately.
Not the test itself, without a license. The distinction that keeps records clean: entering the edition, version, form, score, date, limitations, and interpretation is ordinary result documentation and is fine; building the prompts, stimuli, response rules, or scoring key into an EHR smart form is reproduction and reformatting of protected content, which the publisher says requires written permission, as does electronic administration. Scanned completed protocols sit in between: store them only where the organization is licensed to possess them, with restricted access and audit controls. When in doubt, chart the result and keep the instrument on purchased paper or a licensed electronic platform.
No. Dementia requires clinically established decline plus functional impairment, and the score supplies neither alone. The error runs both directions: a low score in a patient with limited schooling, a different first language, or sensory barriers can be population-typical (the low-education median in US norms sits at 22), while ceiling effects mean a highly educated patient with real impairment can score "normal": in one highly educated cohort the standard cutoff caught barely more than half of dementia cases. The defensible note pairs the number with demographic context, function, and collateral report, and routes genuine concern to fuller evaluation, starting with reversible contributors.
More than the folklore assumes. Published reliable-change estimates run from roughly two to four points in some cohorts to five to seven in others, while average Alzheimer-type decline runs about three points a year, so a small drop is frequently measurement noise. Before interpreting any change, check comparability: same edition and version, same attention method (the two legacy alternatives are not psychometrically equivalent, and a mid-series switch can manufacture "decline"), same language, aids, and clinical state, ideally on alternate forms. Then let function arbitrate: a score falling alongside new help with medications and finances means something a bare number does not.
The 2010 second edition comes in three versions with different totals: Brief (16 points, about five minutes), Standard (30 points, score-equated to the original so longitudinal data carry across), and Expanded (90 points, built to remove the ceiling for subtler impairment), each in Blue and Red alternate forms with published equivalence data. The documentation consequences: never enter a Brief or Expanded total into a classic 30-point severity habit, always name the version and form, and when switching from the original instrument, the Standard Version is the one designed to preserve your serial record.
In Canada and Australia, yes; in the US, no national rule does. Ontario requires 10 to 26 to start and continue cholinesterase-inhibitor coverage; British Columbia uses a Standardised MMSE of 10 to 26 plus a staging scale; Alberta's 2026 criteria accept the MMSE or SLUMS, RUDAS, or interRAI ranges; Manitoba dropped its gate in 2018. Australia's PBS requires a baseline of at least 10 (10 to 14 for memantine), runs a formal low-score pathway when barriers other than disease severity depress the score, and judges continuation on a documented clinically meaningful response, the two-point-improvement rule having been abolished in 2013. For payer submissions, document the administered score with its version and date; a crosswalk estimate never substitutes.
Two different confusions. The mental status exam (MSE) is the descriptive clinical examination of appearance, mood, thought, perception, insight, and judgment; the MMSE is a scored, licensed cognitive screen; a chart can contain either or both, and the shared abbreviation causes real errors. The MoCA is the neighboring screen: markedly more sensitive to mild cognitive impairment (in its validation cohort it detected 90% of MCI against the MMSE's 18%), free to use under its own terms, with its own certification rules. Published crosswalks let you relate MMSE and MoCA scores as estimates, with the original scores always preserved, and no conversion satisfies a payer that names the MMSE.
Yes. Give it the facts (edition, version, form, score, education and language context, prior results, the functional picture) and it drafts the documentation block with the denominator explicit, demographic caveats in place, and the screen-not-diagnosis interpretation scaffolded for your review; it never needs, and should never be given, any test content. It can also check a finished note for a missing version or attention method, a verdict hung on the 24 threshold, or a serial comparison across changed conditions. 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 August 2026:
Educational content, not legal or billing advice. Sample notes are fictional. Follow your organization's policies and your board, payer, and jurisdiction requirements.