Two codes, both at a moderate level of medical decision making. One needs 45 minutes on the time route. The other needs 30.
That is the gap between 99204 and 99214, and almost nothing written about either code says it plainly. A clinician who learned the established patient thresholds and carries them into a new patient visit will reach for 99204 on a 35 minute encounter that does not qualify on time at all.
Here is what the code is, the three year rule that decides whether the patient is new in the first place, how the payment is actually calculated, where the work relative value unit question differs from the payment question, why claims get denied, and a documentation template with two worked examples.
Key Takeaway: CPT code 99204 is the office or other outpatient visit for a new patient requiring a medically appropriate history and examination with a moderate level of medical decision making, or 45 to 59 minutes of total time on the date of the encounter [1]. The decision making bar is the same as 99214. The time bar is not: 99204 asks for 45 minutes where 99214 asks for 30 [1]. What it pays is not one number. Medicare computes the fee schedule amount as the relative value units (RVUs) for the service, multiplied by the geographic adjustment factor (GAF) for your fee schedule area, multiplied by the conversion factor (CF) [2]. RVUs are established separately for physician work, practice expense and malpractice insurance [3], and a non-participating supplier is paid 95 percent of the participating amount [2]. The only current figure for your locality comes from the Centers for Medicare & Medicaid Services (CMS) Physician Fee Schedule Look-Up Tool [4]. Sully.ai drafts the note with the detail that supports the level you actually delivered.
What CPT Code 99204 Is
99204 is the office or other outpatient visit for a newpatient. The descriptor asks for a medically appropriate history and examination, and a moderate level of medical decision making, or 45 to 59 minutes of total time on the date of the encounter [1].
The word carrying the most weight in that sentence is "new". New patient visits are the 99202 to 99205 range. Established patient visits are 99211 to 99215. The two series use the same medical decision making framework and completely different time bands, and which series you are in is settled before any documentation question arises.
New Patient, and the Three Year Rule That Decides It
A patient is new if they have not received a professional service in the past three years from a physician of the same specialty and subspecialty, in the same group practice.
Three conditions, and clinicians routinely remember only the first.
Same specialty and subspecialty. A partner in your group who practices a different specialty does not make the patient established with you. A cardiologist's patient seeing the practice's endocrinologist for the first time is a new patient to the endocrinologist.
Same group practice. A colleague in your own specialty does make the patient established with you, even if you personally have never met them. Covering a partner's panel does not reset the clock.
A professional service, in the past three years. Not a phone call, not a prescription refill handled by staff, but a billable professional service.
Get this wrong and the code is wrong before anything else is considered, and it goes wrong in both directions. Billing a new patient code for an established patient is the error payers catch, and it generates a denial. Billing an established patient code for a genuinely new patient is the error practices never notice, because nothing comes back. It just pays less, quietly, on every misclassified visit.
The Two Routes to 99204
You need one route, not both. Either 45 to 59 minutes of total time on the date of the encounter, or a moderate level of medical decision making [1].
Most visits qualify on decision making. Time is the route for the visit that ran long without especially complex decisions, which happens more often with new patients than with follow-ups because a first visit carries a history that has never been taken.
The full decision making grid, including exactly what moderate requires across the three elements, is covered in the 99214 article rather than repeated here. The bar is the same one.

The Time Threshold That Separates 99204 From 99214
Same Decision Making, Different Clock
Both codes sit at moderate medical decision making. On the decision making route they ask for the same thing.
On the time route they are not close. 99204 is 45 to 59 minutes of total time on the date of the encounter. 99214 is 30 to 39 [1]. The pattern runs through both series: every new patient code asks for substantially more time than the established patient code at the matching decision making level.
The practical consequence is specific. A 35 minute new patient visit does not reach 99204 on time. If the decision making was moderate, the code still stands on the other route and the note has to show it. If the decision making was low, the visit is a 99203, and recording 35 minutes does not change that.
This is the single most common way a new patient visit gets coded on an assumption rather than a rule.
What Counts Toward Total Time
Total time on the date of the encounter, not face to face time. It includes preparing to see the patient and reviewing records beforehand, the visit itself, ordering tests and medications, counseling, documenting in the record, and coordinating care, provided all of it happens on that calendar date [5].
Two rules catch people out.
Only the reporting clinician's time counts. A scribe, a medical assistant or a nurse working on the same visit adds nothing to the clock, however much work they do.
Time on a different calendar date does not count. Reviewing the records the evening before, or finishing the note the next morning, falls outside the encounter date and outside the total.
What 99204 Actually Pays
The leading question people ask about this code is what it charges, and most pages answer with a single national dollar figure. There isn't one.
There Is No National Number, and Here Is Why
Under 42 CFR 414.20, the fee schedule amount for a participating supplier is computed as the product of three things [2]:
The RVUs for the service
The GAF for the fee schedule area
The CF
Two of the three move. The geographic adjustment factor is different in every fee schedule area, so the same code pays differently in Manhattan than in rural Nebraska. The conversion factor is set annually and revised most years. Any figure quoted as "what 99204 pays" is either out of date or correct for exactly one part of the country.
One rule that routinely gets missed: a non-participating supplier is paid 95 percent of the participating amount [2]. That applies to every claim.
The Work RVU Is a Different Question
Two people can ask what a 99204 is worth and mean completely different things, and both are asking something reasonable.
42 CFR 414.22 establishes relative value units separately for three components: physician work, practice expense, and malpractice insurance [3]. The fee schedule amount uses all three. Most employed physician compensation is tied to the workcomponent alone.
So a practice owner billing the visit is looking at the full fee schedule amount for their locality. An employed physician on a productivity contract is looking at the work RVU, which is a different number measuring a different thing. Neither is wrong, and conflating them produces arguments that cannot be settled because the two sides are quoting incompatible figures.
One structural difference is worth knowing. The work RVU does not change with your locality. The geographic adjustment factor does, and the practice expense component is lower when you provide the service in a facility than in an office, because the facility carries the overhead. Same work, same code, different payment, depending on where the visit happened. The physician's work RVU is unaffected by any of that.
Both sets of values are revised annually, which is why this article does not print either one.

How to Get Your Own Number
Use the CMS Physician Fee Schedule Look-Up Tool [4]. Four choices and you have the real figures:
Select the year you are billing for
Enter 99204
Select your Medicare Administrative Contractor (MAC) locality
Choose facility or non-facility
That returns the fee schedule amount for your area and the relative value unit components behind it, which answers both versions of the question in one place.
One caveat worth stating.That gives you the Medicare amount. Commercial payers set their own rates, frequently as a percentage of the Medicare schedule, sometimes well above or below it. Treat the Medicare figure as your reference point rather than your full picture.
99203 or 99204
Both are new patient visits, so the three year rule is already settled. What separates them is the same thing that separates 99213 from 99214, applied to the new patient series.
99203 is a low level of medical decision making, or 30 to 44 minutes of total time. 99204 is moderate, or 45 to 59 [1][5].
On decision making, the move from low to moderate is usually one element rather than a different kind of visit. A second stable chronic problem addressed rather than one. Prescription drug management, which reaches moderate risk on its own. The low end of that grid is laid out in the 99213 article.
On time, the gap is narrower than people expect. A 44 minute visit is a 99203. A 45 minute visit is a 99204. One minute of recorded total time separates them, which is the strongest practical argument for writing down the actual figure instead of a rounded guess. "About 45 minutes" is not a time entry a reviewer can use.
Why a 99204 Gets Denied

The patient was not new. Someone of your specialty and subspecialty, in your group practice, provided a professional service inside the past three years. The established patient code applies instead, and the three year window is the first thing a payer checks.
Neither route is evidenced. The note records no total time on the date of the encounter, and it does not establish a moderate level of medical decision making. Without one of the two, nothing supports the level.
Time was recorded but not qualified. A note that says "45 minutes" without saying it is total time on the date of the encounter, or a note where the figure includes staff time, does not support the time route. The number on its own is not the evidence. What the number measures is.
The diagnosis does not support medical necessity. The visit can be documented perfectly and still be denied if the diagnosis submitted does not justify a new patient visit at that level.
The 99204 Documentation Template
One block, complete as written. Paste it straight into a chart, or start from the progress note template and add the new patient lines.
OFFICE VISIT, NEW PATIENT
Patient, date of service: (and the rendering clinician)
New patient confirmation: (no professional service from a same-specialty clinician in this group within three years, with the date checked)
Referral source: (who sent them, and the question being asked)
Chief complaint: "(in the patient's words)"
History: (the full history, since none of it is on file yet)
Examination: (findings relevant to the problems being addressed, with values)
Problems addressed today: (each one named, with its status: stable, worsening, new)
Data reviewed: (outside records, labs, imaging, notes from other clinicians, or "none" if none)
Risk of management selected: (what you decided, and what the risk of that decision is)
Route used: (time, or medical decision making; state which)
If time: (total time on the date of the encounter, stated as total time, and what it comprised)
Assessment and plan, per problem: (each problem gets its own assessment and its own plan, the same structure a SOAP note uses)
Signature, credentials, date: (the clinician who performed the service)
Three lines carry the level here rather than two. "Problems addressed today" and "Risk of management selected" decide the decision making route. "New patient confirmation" decides whether you are in the right code series at all, and it is the line almost no template includes.
Where Sully.ai Fits in a New Patient Visit
A first visit covers more ground than a follow-up. A history that has never been taken, outside records read for the first time, several problems named in one sitting, and the reasoning behind each decision spoken aloud in the room.
All of that has to survive into a note usually written three patients later. What gets lost is the specific detail that evidences the level: the second problem addressed briefly, the outside record you read before walking in, the reasoning behind starting a medication rather than waiting.
Sully.ai's AI scribe drafts the note from the visit, so the problems addressed, the data reviewed and the risk are on the page rather than reconstructed later. It runs on a single integration across Epic, Cerner, Meditech and Athenahealth.
Two boundaries, and both are refusals.
Nothing should choose the code. Code selection is your attestation, it is what an auditor examines, and it carries liability belonging to the person who saw the patient. A tool that picks the level and asks you to approve it has moved the decision without moving the responsibility.
No documentation tool adds to your clock. Only the reporting clinician's own time counts toward the time route. A scribe of any kind, working alongside you on the same visit, contributes nothing to the 45 minutes. This matters more for new patient codes than for established ones, because the time route is reached more often on a first visit.
Sully operates across 5,000+ providers, has delivered 50M+ hours of work, and prices each role 80 to 90 percent below the human equivalent [6].
Two Worked Examples
A New Patient Visit That Is Clearly 99204
New patient, referred by a nurse practitioner at an urgent care. No prior service from anyone in this group. 51 minutes total time on the date of the encounter.
New patient confirmation: No professional service from any internal medicine clinician in this group. Chart search returns no prior encounters.
Referral source: Urgent care, 12 September, for persistently elevated blood pressure and an incidental finding of elevated glucose.
Chief complaint: "They told me my blood pressure was high and my sugar was off, and I should get a real doctor."
History: 54 year old man, no established primary care in over a decade. Hypertension never formally diagnosed, no treatment. Father and brother both with type 2 diabetes. Smokes half a pack daily for 30 years. No chest pain, no shortness of breath on exertion, no visual change, no polyuria. Not on any prescription medication.
Examination: Blood pressure 158/96, repeated 154/94. Heart rate 78 regular. Body mass index 32. Chest clear. Feet examined, sensation intact, no ulceration. No retinopathy on fundoscopy.
Problems addressed today: Essential hypertension, newly diagnosed and untreated. Type 2 diabetes, newly diagnosed. Tobacco use disorder.
Data reviewed: Urgent care record from 12 September obtained and read, including blood pressure readings and a random glucose of 232 mg/dL. HbA1c drawn today, 8.1 percent, resulted during the visit. Basic metabolic panel and lipid panel reviewed.
Risk of management selected: Starting two new prescriptions. Prescription drug management, moderate risk.
Route used: Medical decision making. Time also documented.
Total time on the date of the encounter: 51 minutes, comprising review of the outside urgent care record before the visit, the encounter itself, ordering and reviewing labs, counseling on diagnosis and smoking, and documentation.
Assessment and plan: Hypertension, newly diagnosed, stage 2. Starting lisinopril 10mg daily. Type 2 diabetes, newly diagnosed, confirmed on HbA1c. Starting metformin 500mg twice daily with food, counseled on gastrointestinal effects and titration. Tobacco use disorder, discussed cessation, patient not ready to set a date, will revisit. Repeat basic metabolic panel in 2 weeks for renal function. Review in 4 weeks. HbA1c in 3 months.
Three problems addressed, two of them new. An external record obtained and reviewed, plus independently interpreted labs. Two new prescriptions started. Moderate decision making across all three elements, which is more than the two needed. The visit also clears 45 minutes, so both routes support the code. The note states which one was used.
A Visit That Looks Like 99204 and Is Actually 99203
Same clinic, a different new patient the same week. 40 minutes total time.
New patient confirmation: No professional service from any internal medicine clinician in this group within three years. Last seen by this group's orthopedic surgeon in 2024, different specialty, so the patient is new to internal medicine.
Referral source: Self referred, establishing care after moving to the area.
Chief complaint: "I just need someone local to keep an eye on my thyroid."
History: 41 year old woman, hypothyroidism diagnosed 2018, stable on levothyroxine 75mcg daily for the past four years. No fatigue, no weight change, no cold intolerance, no palpitations. Takes the medication consistently on an empty stomach. No other medical history. No other medications.
Examination: Blood pressure 118/74. Heart rate 68 regular. Thyroid not enlarged, no nodules palpable. No tremor. Reflexes normal.
Problems addressed today: Hypothyroidism, stable.
Data reviewed: Thyroid stimulating hormone from her previous practice, 2.1 four months ago, within range. No new labs today.
Risk of management selected: Continue current dose unchanged. Low risk.
Route used: Medical decision making.
Total time on the date of the encounter: 40 minutes, comprising records transfer review, the encounter, and documentation.
Assessment and plan: Hypothyroidism, well controlled on current dose. Continue levothyroxine 75mcg daily. Repeat thyroid stimulating hormone in 6 months. Establish care, routine health maintenance discussed and scheduled separately. Review in 6 months.
Why this is a 99203. One stable chronic illness is low under the problems element. Minimal data, none of it newly interpreted. Continuing an unchanged medication is not prescription drug management, so risk stays low. That is low decision making, not moderate.
And the time route does not rescue it. 40 minutes falls inside the 99203 band of 30 to 44 minutes, not the 99204 band of 45 to 59 [1]. The visit took real time, and it was still a smaller visit.
Note what the second example is not. It is not a badly documented 99204. It is a well documented 99203, and the note makes that case clearly enough that nobody has to guess. That is what both codes are supposed to look like.
FAQ
What is CPT code 99204?
The office or other outpatient visit for a new patient requiring a medically appropriate history and examination with a moderate level of medical decision making, or 45 to 59 minutes of total time on the date of the encounter [1].
What is the charge for CPT code 99204?
There is no single figure. Medicare computes the fee schedule amount as the RVUs for the service multiplied by the geographic adjustment factor for your area and the conversion factor [2], so it varies by locality and by year. A non-participating supplier is paid 95 percent of the participating amount [2]. Get the current number for your locality from the CMS Physician Fee Schedule Look-Up Tool [4]. Commercial payers set their own rates separately.
What is the difference between CPT 99204 and 99214?
The patient, and the clock. 99204 is a new patient visit and 99214 is an established patient visit. Both sit at a moderate level of medical decision making, so on that route they ask for the same thing. On the time route they do not: 99204 needs 45 to 59 minutes where 99214 needs 30 to 39 [1].
What is the difference between CPT code 99204 and 99203?
Both are new patient visits. 99203 is a low level of medical decision making, or 30 to 44 minutes of total time. 99204 is moderate, or 45 to 59 [1][5].
Is CPT code 99204 still valid?
Yes. The 2021 revision to office visit coding changed how the level is determined, removing history and examination as level-setting elements and replacing them with medical decision making or total time [5]. The code itself was not retired, and the confusion comes from guidance published before that change.
Sources
[1] American Medical Association — CPT code 99204, new patient office visit, 45-59 minutes. https://www.ama-assn.org/practice-management/cpt/cpt-code-99204-new-patient-office-visit-45-59-minutes
[2] U.S. Government Publishing Office — 42 CFR 414.20, Formula for computing fee schedule amounts. https://www.govinfo.gov/content/pkg/CFR-2023-title42-vol3/xml/CFR-2023-title42-vol3-sec414-20.xml
[3] U.S. Government Publishing Office — 42 CFR 414.22, Relative value units (RVUs). https://www.govinfo.gov/content/pkg/CFR-2023-title42-vol3/xml/CFR-2023-title42-vol3-sec414-22.xml
[4] Centers for Medicare & Medicaid Services — Physician Fee Schedule Look-Up Tool. https://www.cms.gov/medicare/physician-fee-schedule/search
[5] Centers for Medicare & Medicaid Services — Evaluation and Management Services, MLN006764. https://www.cms.gov/outreach-and-education/medicare-learning-network-mln/mlnproducts/downloads/eval-mgmt-serv-guide-icn006764.pdf
[6] Sully.ai — AI Medical Scribe. https://www.sully.ai/medical-scribe
TABLE OF CONTENTS
Hire your
Medical AI Team
Take a look at our Medical AI Team
AI Receptionist
Manages patient scheduling, communications, and front-desk operations across all channels.
AI Scribe
Documents clinical encounters and maintains accurate EHR/EMR records in real-time.
AI Medical Coder
Assigns and validates medical codes to ensure accurate billing and regulatory compliance.
AI Nurse
Assesses patient urgency and coordinates appropriate care pathways based on clinical needs.