Sign Up
Login
Try for Free
Login
The MoCA (Montreal Cognitive Assessment) is a 30-point cognitive screening instrument developed by Ziad Nasreddine and validated in 2005, built to detect mild cognitive impairment. Primary care, neurology, and geriatrics teams use it at wellness visits and in memory workups. This page covers how to document and interpret MoCA results in the chart, with a fictional sample note.
Any healthcare professional for total-score screening; certification required to interpret subscores and domains (policy dated December 2025)
Referring and treating clinicians, memory clinics and neuropsychologists, payers and care programs, families receiving results
3 to 8 chart lines · administration about 10 minutes
Clinician-administered cognitive screen (30 points; adapted versions scored out of 22 and 15)
Medicare annual wellness visits, memory complaints and MCI workups, delirium recovery tracking, pre-treatment cognitive baselines
Accessed via mocacognition.com; certification optional for total-score screening since December 2025; no authority mandates it by name
The Montreal Cognitive Assessment (MoCA) is a roughly ten-minute, 30-point cognitive screening instrument developed by neurologist Ziad Nasreddine and validated in 2005 as a tool sensitive to mild cognitive impairment. It is a trademarked instrument accessed through mocacognition.com, and its rules have moved: mandatory universal training, announced in 2019, gave way in a policy dated December 2025 to a tiered model in which certification is optional for healthcare professionals who use only the total score for screening, triage, or referral, and required for anyone interpreting individual tasks, subscores, or domain-specific impairments, with exemptions for neuropsychologists and clinicians with a year of postdoctoral cognitive-assessment training. Training is free for students, faculty, academic researchers, and full-time employees of publicly operated healthcare institutions; the official site no longer publishes a standard paid price for everyone else. One caution the site itself creates: older Terms and Permissions pages still carry universal-certification language, so institutions relying on the exemption should keep a copy of the dated FAQ or obtain written confirmation.
The load-bearing documentation facts are arithmetic. The Full versions (8.1, with alternate forms 8.2 and 8.3 for repeat testing) are scored out of 30, with one education correction: add one point for 12 or fewer completed years of education, only when the total is below 30, and chart the raw and adjusted values separately. The version family changes the denominator: MoCA-Basic serves low-literacy populations at /30, the Blind/Telephone version is scored out of 22, and the five-minute version out of 15, and the official FAQ states plainly that converting a /22 result to /30 is not validated. And the famous cutoff of 26 is a sensitivity-oriented threshold from the 2005 validation sample, not a verdict: in a diverse population-based sample, 66% of adults scored below it, and meta-analytic work supports lower cutoffs near 23 in many settings. A MoCA result is a screen that feeds a fuller evaluation; it diagnoses nothing by itself.
Primary care teams document MoCA results at Medicare annual wellness visits, where the law requires detection of cognitive impairment but names no instrument; clinicians billing cognitive assessment and care plan services fold it into a much larger documentation set (independent historian, function, safety, staging instruments, care plan); memory clinics use it as the brief baseline before neuropsychological evaluation and treatment pathways, where payer rules sometimes accept a MoCA range as an eligibility gate; hospital teams track it through delirium recovery, with the during-delirium caveat written in; and capacity and driving questions cite it only as collateral evidence, never as the determinant, alongside a capacity evaluation. Neighbors matter: the MMSE is a licensed Pearson-era incumbent that is less sensitive to MCI, SLUMS and Mini-Cog are free alternatives, RUDAS serves culturally and linguistically diverse populations, and the RBANS is the purchased next step when domain profiling with repeatable forms is needed.
No statute prescribes a MoCA note format. The elements below are the convention that keeps a screening number interpretable months later, when a different clinician compares against it. Each carries the pitfall that most often undermines it.
Clinical state and purpose. Say why the screen was done and the patient's state: routine screening, a complaint, a baseline before treatment, or recovery tracking, and whether delirium, acute illness, or major fatigue was present. Pitfall: a score obtained during delirium charted without that context, then treated as the premorbid baseline forever after.
Exact version, language, mode, and denominator. Name the form (Full v8.1, 8.2, or 8.3, Basic, Blind/Telephone, five-minute, hearing-adapted, digital), the language, the mode (in person, telephone, videoconference), and the denominator. Pitfall: "MoCA 18" from a /22 telephone version read against the /30 cutoff. The official FAQ states that converting /22 results, or scores from partially administered Full versions, to /30 is not validated.
Raw and adjusted scores, both. Record the raw total, the completed years of education, and whether the one-point Full-version correction was applied (12 or fewer years, only when the total is below 30). Pitfall: one merged number. A raw 24 with correction and an uncorrected 25 look identical downstream, and serial comparison quietly breaks; correction applied at 30/30 or at every education level is simply wrong.
Testing conditions. Note hearing aids, corrective lenses, motor limits, literacy, interpreter use, interruptions, and anything that deviated from standard administration. Pitfall: conditions omitted. A later three-point difference cannot be interpreted when nobody recorded that the baseline was done without hearing aids in a noisy room.
Domain observations, only when qualified, as patterns. Under the December 2025 policy, interpreting subscores and domains requires certification or an exemption; where qualified, chart relative patterns ("relative weakness in new learning and executive control") rather than localization. Pitfall: mapping a missed task to an etiology. Large registry analyses find domain scores substantially less reliable than the total.
Interpretation with the cutoff in context. Below 26 means below the original sensitivity-oriented validation threshold, a flag for further evaluation; it is not "abnormal, has dementia." Meta-analytic work supports cutoffs near 23 in many settings, and a high score does not rule impairment out. Pitfall: verdict language. In one diverse population sample, 66% scored below 26; among cognitively healthy adults aged 70 to 99, 46% did.
Function, plan, and serial comparability. Tie the score to daily function and collateral report, state the plan (reversible-cause workup, monitoring, referral), and set up the next administration: same language and mode, an alternate form at intervals of about three months or less, and the exact form recorded. Pitfall: over-reading small changes. Published reliable-change estimates run from roughly two to four points depending on cohort and interval, so a one- or two-point movement is usually measurement noise unless function moves with it.
MoCA DOCUMENTATION BLOCK Reason / clinical state: [screening / complaint / baseline / recovery tracking; delirium or acute illness present?] Version: [Full v8.1 / 8.2 / 8.3 / Basic / Blind-Telephone / 5-minute / other] Language: [ ] Mode: [ ] Date: [ ] Administered by: [ ] Raw score: [ ]/[30 / 22 / 15] Education: [ ] completed years Correction: [+1 applied / not applied / n/a] Adjusted: [ ]/30 Conditions: [hearing aids / lenses / motor limits / interpreter / interruptions / deviations] Domain observations (certified or exempt only): [relative patterns, not localization] Interpretation: [screening result vs applicable threshold in this population; not a stand-alone diagnosis] Function / collateral: [IADLs, medication and finance management, informant report] Plan: [reversible workup / monitoring / referral / none] Next administration: [interval; alternate form if under ~3 months; same language and mode] 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.
Scenario: an annual wellness visit with a borderline result and a collateral concern, the everyday case where the difference between a defensible note and "MoCA 24" matters most. All details are fictional.
Patient: H.B., 74 · Visit: Annual wellness visit · Clinician: S. Okafor, MD (family medicine) · Note date: 08/10/2026
Context: Cognitive screening at the annual wellness visit. H.B. reports occasional word-finding pauses; her husband, present with consent, describes two medication-refill errors in the past three months. No delirium, acute illness, or sleep deprivation today.
Measure: MoCA Full v8.1, English, in person, administered by this clinician: raw 24/30. Education 12 completed years; one-point Full-version correction applied; adjusted 25/30. Hearing aids worn, reading glasses used, no interruptions; standard administration throughout. Total-score screening use only; no domain interpretation is charted.
Interpretation: The adjusted score falls below the original validation threshold of 26, a screening flag rather than a diagnosis, and population studies show many cognitively healthy older adults score in this range. Taken with the collateral medication-management concern, the result supports further evaluation; it does not establish mild cognitive impairment or dementia.
Function and plan: Basic daily activities intact; instrumental activities notable for the refill errors. Plan: medication reconciliation with a simplified regimen, laboratory review of reversible contributors, mood and sleep screening, and a repeat administration in about 12 months using alternate form 8.2 under the same conditions, sooner if function changes. Referral for fuller cognitive evaluation will follow if the concern persists or the trajectory declines. Results discussed with H.B. and her husband in plain language: areas worth watching, nothing diagnosed today.
This sample is fictional and for educational purposes. It does not describe a real patient or record, 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 bulletsIn the US, the rules around cognitive screening are mostly tool-neutral, and the note should reflect which layer it is serving. The Medicare annual wellness visit regulation (42 CFR 410.15) is LAW that requires detection of cognitive impairment through direct observation and patient and informant report; it names no instrument, so "Medicare requires a MoCA" is false, while a MoCA can document the detection work. The cognitive assessment and care plan service (99483) is PAYER POLICY that expects far more than a score: an independent historian, function and decision-making ability, behavioral symptoms, safety including driving, caregiver needs, staging instruments, and a written care plan; "MoCA 20/30" alone supports nothing, and some local contractor articles name a short MoCA as one acceptable documentation route, which is a local rule rather than a national mandate. CMS's GUIDE dementia model runs its tiering on the CDR, FAST, and Zarit instruments, not the MoCA. Treatment gates deserve their own label: Medicare's national anti-amyloid coverage framework sets no numerical MoCA threshold, while some commercial policies (for example, 2026 UnitedHealthcare criteria) accept a MoCA of 17 to 30 as one eligibility route; those ranges are payer gates, not clinical cutoffs or validated staging. And for the two highest-stakes questions, the score is collateral evidence only: capacity is decision-specific and functional, documented in a capacity evaluation, and driving-fitness guidance (NHTSA and the AGS in the US) treats cognitive screens as triggers for fuller evaluation, with state reporting laws that turn on clinical judgment rather than any score.
Canada and Australia run the same logic with their own anchors. The Canadian Medical Association's 2025 driver's guide states that a cognitive screen generally cannot determine fitness to drive by itself except in severe impairment, provincial reporting duties (mandatory in some provinces, discretionary in others) turn on the clinician's overall judgment, and provincial drug plans historically key coverage to MMSE-family scores rather than the MoCA (British Columbia's donepezil criteria use the SMMSE), while Canada's August 2026 national lecanemab recommendation permits clinically accepted alternatives such as the MoCA without setting a numerical threshold. In Australia, Austroads' fitness-to-drive framework sets no MoCA cutoff, the June 2026 aged-care assessment materials name GPCOG and KICA pathways rather than the MoCA, RUDAS remains the culturally responsive alternative for diverse and low-education populations, and current PBS dementia-drug criteria publish no MoCA threshold. Across all three countries, the measurement cautions are the same: nearly 100 translations exist but availability is not validation, remote administrations agree only moderately with in-person testing at the individual level, the digital product line is in transition (MoCA Duo is being retired for MoCA Solo during 2026, and the public XpressO self-test is an awareness tool, not a clinical administration), so the chart names the exact product, language, and mode every time.
MoCA is a registered trademark of TEST MOCA INC., and the instrument is distributed through mocacognition.com under its permission terms, which prohibit republishing the form; this page links to the official source and hosts nothing. BastionGPT is not affiliated with, or endorsed by, MoCA Cognition, TEST MOCA INC., or the instrument's developers. This page reproduces no test items, stimuli, forms, scoring keys, or normative tables.
The over-identification data are the backdrop for most of these. In a diverse population-based sample of 2,653 adults, 66% scored below the conventional cutoff of 26; among cognitively healthy adults aged 70 to 99, 46% did; a Cochrane review found more than 40% of people without dementia screen positive at that threshold; and meta-analytic work supports cutoffs near 23 in many settings. Meanwhile the instrument's own rules moved in December 2025, and most template pages have not noticed. The BastionGPT Clinical Advisory Board sees the same errors most often in MoCA 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.
The Full versions total 30 points, and the familiar threshold of 26 comes from the 2005 validation study, where it caught 90% of mild cognitive impairment and 100% of mild Alzheimer disease in a clinically enriched sample. In broader populations it over-identifies: 66% of a diverse community sample and 46% of cognitively healthy adults aged 70 to 99 scored below 26, and meta-analyses support cutoffs near 23 in many settings. So a below-threshold score is a flag for evaluation of reversible causes, function, and, where indicated, fuller testing; it is not a diagnosis, and a high score does not rule impairment out.
Under the policy dated December 2025 on the official MoCA FAQ: no, when a healthcare professional uses only the total score for screening, triage, or referral; yes, when anyone interprets individual tasks, subscores, or domain-specific impairment, with exemptions for neuropsychologists and clinicians with a year of postdoctoral cognitive-assessment training. Training is free for students, faculty, academic researchers, and employees of publicly operated healthcare institutions. Two cautions: older pages on the same site still carry the 2019-era universal-mandate language, so keep the dated FAQ or written confirmation, and your employer may require training regardless.
For the Full versions: add one point when the patient has 12 or fewer completed years of formal education (counted after kindergarten), only when the total is below 30, and never above 30. Chart the raw score, the education years, and the correction status as three separate facts: "raw 24/30; 12 years education; +1 applied; adjusted 25/30." A single merged number breaks serial comparison, because nobody can later tell whether a one-point difference is clinical or bookkeeping. And the correction does not fully remove education effects: research finds lower-education patients remain more likely to be misclassified even after the point is applied, which belongs in the interpretation, not in improvised extra points.
By their own names and denominators. The Blind/Telephone version is scored out of 22, the five-minute version out of 15, and MoCA-Basic (for low-literacy populations) out of 30 as its own adapted instrument. The official FAQ states that converting a /22 result to /30, or prorating a Full administration with omitted visual items, is not validated, so write "MoCA Blind/Telephone 18/22," never "MoCA 18" or "equivalent to 25/30." If a data system forces a /30 field, record the native result in text and label any system-generated conversion as unvalidated. Telephone adaptations in the literature vary in content and maxima, which is exactly why the note names the specific version.
No. A dementia diagnosis requires evidence of cognitive decline plus functional impairment established clinically; a screening score supplies neither by itself, and the official MoCA materials say a low result does not establish a diagnosis while portions of the published severity bands were set arbitrarily. The defensible note reads: below the applicable threshold, screening result, reversible contributors reviewed (medications, mood, sleep, sensory, metabolic), function and collateral documented, and a plan: monitoring, repeat testing under comparable conditions, or referral for neuropsychological evaluation when the stakes or the uncertainty warrant it.
More than most notes assume. Published reliable-change estimates run from roughly two points in some cohorts to about four in others, practice effects inflate early retests, and the alternate forms (8.2 and 8.3), built for repeat testing at intervals of about three months or less, are close but not perfectly matched in difficulty. So a one- or two-point movement is usually within measurement error, and even a three- or four-point change deserves a check: same form lineage, language, mode, sensory aids, and clinical state? The strongest evidence of real change is concordance, a score moving in the same direction as function and collateral report, and that is what the note should show.
No, in every jurisdiction this page covers. Capacity is decision-specific: the conclusion rests on the patient's demonstrated ability to understand, appreciate, reason about, and communicate a specific choice, documented in a capacity evaluation, with the MoCA as collateral cognitive evidence only. Driving guidance runs the same way: US clinician guidance, the Canadian Medical Association's 2025 driver's guide, and Australia's Austroads framework all treat cognitive screens as triggers for fuller assessment (collateral history, occupational-therapy or on-road evaluation), not as pass/fail gates, and state and provincial reporting laws turn on clinical judgment about the condition, not on a score.
By purpose, population, and licensing. The MoCA is more sensitive to mild cognitive impairment than the MMSE (which is a licensed, purchased instrument with its own documentation rules) at the cost of more false positives in low-prevalence and low-education settings. SLUMS is a free alternative for older adults, Mini-Cog is the very brief triage option, and RUDAS was built for culturally and linguistically diverse and low-education populations, where the MoCA's conventional threshold misfires most. When the question outgrows any screen (domain profiling, validity assessment, high-stakes decisions), the next steps are an RBANS or a full neuropsychological evaluation. Never convert scores between instruments; chart each by name.
Yes. Give it the facts (version, raw total, education years, conditions, prior scores, the functional picture) and it drafts the documentation block: raw and adjusted values separated, the screening-not-diagnosis caveat in place, and the plan and serial-comparison sentences scaffolded for your review. It can also check a finished note for a denominator mismatch, a misapplied education correction, verdict language around the cutoff, 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.