CPT 90791: the code that sets up everything after it

90791 is a psychiatric diagnostic evaluation without medical services: a biopsychosocial assessment, history, mental status examination, diagnostic formulation and initial treatment recommendations. Where medical services are part of the evaluation — prescribing, physical examination, medical decision-making — the code is 90792 instead, and only prescribers may bill it.

What actually has to be in it

90791 has no time band, which surprises people. It is not a timed code. What defines it is scope, and payers assess scope through the components present in the note.

• Presenting problem and history of the current episode.
• Relevant psychiatric, medical, family and social history, including substance use.
• Mental status examination.
• Risk assessment — suicidality, homicidality, self-harm, and disposition where risk is present. This is the component most often thin and the one most costly to omit.
• Diagnostic formulation with the reasoning that supports the diagnosis, not just the code.
• Initial treatment recommendations — the seed of the treatment plan.

The formulation is where evaluations most often fall short. A note listing F41.1 without articulating why the presentation meets that threshold rather than an adjacent one has recorded a conclusion rather than an evaluation, and every later claim inherits that weakness. See the golden thread.

Who can bill it

90791 is available to qualified behavioral health clinicians operating within their licensure and scope — psychologists, clinical social workers, counsellors and marriage and family therapists, alongside prescribers. Which of those a given payer credentials and reimburses is a separate question, and one that varies considerably by plan and state, particularly for Medicaid and for licensure categories that reached independent Medicare billing more recently.

Frequency, and the duplicate-denial problem

This is the practical heart of the code. Most payers reimburse one 90791 per episode of care, commonly with a look-back window measured in months or a year. Beyond that, expect denial as a duplicate unless a re-evaluation is justified.

The denials that frustrate practices usually come from one of four situations, none of which is clinically unreasonable:

1. Transfer within a practice. A client moves from one clinician to another and the new clinician performs their own evaluation. Clinically appropriate; frequently denied as a duplicate because the payer sees the same tax ID and the same client within the window.
2. Return after a gap. A client returns after eight months. Whether that is a new episode of care or a continuation is a payer-policy question, and the policies differ.
3. Concurrent evaluations. A therapist bills 90791 and a prescriber bills 90792 in the same window. Some payers permit this; others reject the second as duplicative.
4. Significant clinical change. A materially different presentation warranting full re-evaluation. Usually defensible, but only if the note says so explicitly.

What it is not

Not a first therapy session

An intake that consists mostly of rapport-building and beginning treatment is a psychotherapy session, not a diagnostic evaluation. If the components above are not present, the honest code is 90834 or 90837.

Not billable alongside psychotherapy on the same day, usually

Most payers do not reimburse 90791 and an individual psychotherapy code on the same date for the same client. Where an evaluation and a substantive therapy session genuinely both occurred, this needs checking against policy rather than assuming.

Not a form

An intake questionnaire completed by the client, however thorough, is not a diagnostic evaluation. The evaluation is the clinician's assessment, formulation and reasoning. Structured instruments support it; they do not constitute it.

Telehealth

90791 is widely reimbursed by telehealth, with the usual caveats: place-of-service and modifier requirements vary by payer and have changed repeatedly since 2020. Audio-only evaluations are a narrower and more variable case — some payers permit them, some restrict them by population or geography, and some do not cover them at all.

What Weft does with this

Weft treats the evaluation as the origin of the thread rather than a form to complete. The diagnosis it establishes populates the treatment plan; the plan's goals are what later notes are written against; and if the diagnosis is later refined, the plan surfaces for review rather than silently diverging from the claims. The intake stops being a document you file and becomes the structure everything downstream inherits.

The risk assessment specifically

Of all the components, this is the one worth writing carefully, and not primarily for billing reasons. A risk assessment that records a conclusion without the findings behind it — "denies SI" as the entire entry — is thin in the exact circumstance where thinness carries the most consequence, clinically and legally.

A defensible entry records what was asked and what was found: ideation and its character, intent, plan, means and access to them, protective factors, history of attempts, and the reasoning that connects those findings to the disposition. Where risk is present, the plan follows — safety planning, means restriction, increased contact, referral — with what was agreed and with whom.

This is the section that matters if a case is ever reviewed by a board or a court, where the standard is not whether the outcome was good but whether the assessment and response were reasonable and documented at the time.

Intake in a group practice

Group practices hit two structural problems with 90791 that solo practices do not.

The first is the transfer duplicate described above: a client reassigned between clinicians generates a second evaluation under the same tax ID, which payers frequently reject. Some practices resolve this by having the receiving clinician document a transfer-of-care review rather than a full re-evaluation, billing a psychotherapy code for the session. That is often the honest description of what actually happened.

The second is intake performed by one person and treatment by another — an intake coordinator or assessment clinician evaluates, then hands off. This is operationally efficient and creates a documentation gap if the treating clinician never records their own agreement with the formulation. The treatment plan should show the treating clinician adopting or revising the diagnosis, not silently inheriting it.

Reassessment cadence

A question 90791 raises but does not answer: how often should the diagnosis and formulation be revisited during a long episode of care?

There is no CPT answer, because reassessment is usually not a separately billable event — it is documentation within an ordinary session. But payers reviewing long-running treatment look for evidence that someone checked whether the original formulation still fits, and treatment plans in most jurisdictions carry their own review requirements, commonly every 90 days to a year depending on setting and payer.

A short documented review at a regular interval — diagnosis still accurate, goals still current, measures trending, plan revised or continued with reasoning — closes the most common gap in long-episode charts and takes a few minutes.

Verified 29 July 2026 against AMA CPT descriptors and CMS published guidance. Code definitions are national; coverage, reimbursement and documentation expectations are set by individual payers and vary by plan, state and contract. Primary references: AMA CPT; CMS Physician Fee Schedule; APA Services. This page is billing reference, not legal or coding advice.