Radiation delivery coding stopped being about IMRT vs 3D on January 1, 2026. The fight now is whether your chart proves Level 3, and whether you tell CMS what 2027 payment should look like before comments close.

Introduction

On January 1, 2026, Medicare stopped paying freestanding and hospital outpatient radiation delivery the old way. Technique-specific IMRT codes and the long-running HCPCS G6001, G6017 series were deleted. In their place sit three revised, technique-agnostic CPT codes, 77402 (Level 1), 77407 (Level 2), and 77412 (Level 3), with imaging guidance technical services bundled into delivery and the professional component of image guidance consolidated into 77387.

CMS projected something close to specialty-level budget neutrality. ASTRO’s national survey, ASCO’s April 2026 practice brief, and clinic-level revenue reports tell a different story: many practices saw drops of 10% or more, Level 3 claims are frequently reclassified or denied, and commercial and Medicaid systems are still catching up to the January rewrite.

Here is the main point: Level 3 is not a technique badge and not a payer preference. It is a documentation and prior-authorization problem with a hard coding definition. Get the definition wrong, or document “DIBH” as a checkbox, and 77412 becomes 77407 before the first fraction is paid. At the same time, CMS is already reopening radiation delivery valuation for CY 2027 in proposed rule CMS-1848-P, with comments due September 14, 2026. Independent practices that only fix 2026 claim edits and ignore the 2027 comment window will fight the same payment fight again next year.

This post walks through the code family, the Level 3 criteria, the denial patterns ASTRO has published, the documentation stack that actually survives review, the related 77387 professional-component mess, a practical workflow for independent clinics, common failure modes, and what to put in a CMS comment before the September 14 deadline.

Background: why the 2026 rewrite happened

For more than a decade, freestanding centers and hospital outpatient departments lived in different coding languages for the same megavoltage delivery. Hospitals used CPT delivery codes. Freestanding centers used G-codes that CMS created in part to preserve technical payment when earlier packaging rules collided with how radiation was billed. IMRT had its own delivery pair (77385 / 77386). CT guidance for field placement (77014) and various imaging G-codes sat alongside delivery.

That structure no longer matched how clinics treat patients. Modern external-beam programs mix 3D conformal, IMRT/VMAT, surface guidance, gating, breath-hold, multi-isocenter plans, mixed electron/photon setups, and total skin electrons, often in the same department on the same day. The AMA CPT Editorial Panel, with specialty society input, approved a complexity-based family rather than a technique-based one. CMS implemented that family across settings effective January 1, 2026.

CMS MLN Matters MM14361 (Hospital Outpatient Prospective Payment System: January 2026 Update) is explicit for non-excepted off-campus provider-based departments: G6001, G6017 are deleted; bill 77402 / 77407 / 77412 with modifier PN so those departments continue to receive the PFS-equivalent rate. Do not resurrect the G-codes for dates of service on or after January 1, 2026.

ASCO’s April 7, 2026 practice analysis summarized the operational facts cleanly:

• Delivery hierarchy: 77402 Level 1, 77407 Level 2, 77412 Level 3
• Technique no longer drives the code; complexity does (multiple isocenters, active motion management, and the other Level 3 descriptors)
• Technical image guidance is bundled into delivery
• Professional image guidance consolidates to 77387
• Deleted: 77385, 77386, 77014, and G6001, G6017

ASTRO’s Winter 2026 ASTROnews published the same descriptors and stressed the definitional pivot: the codes “no longer distinguish between 3D and IMRT delivery and instead are based on the complexity of the treatment delivery itself.”

That sentence is the whole ballgame. If your EHR still maps “IMRT = high code,” you are coding 2015 logic in a 2026 world.

The code map, with ASTRO’s official wording

Use the ASTRO Winter 2026 descriptors as your staff training source of truth (AMA CPT owns the copyrighted language; ASTRO republished the clinical framing practices are using day to day):

Level 1, CPT 77402

Radiation treatment delivery; Level 1 (for example, single electron field, multiple electron fields, or 2D photons), including imaging guidance, when performed.

Level 2, CPT 77407

Radiation treatment delivery; Level 2, single isocenter (e.g., 3D or IMRT), photons, including imaging guidance, when performed.

Level 3, CPT 77412

Radiation treatment delivery; Level 3, multiple isocenters with photon therapy (for example, 2D, 3D, or IMRT) OR a single isocenter photon therapy (e.g., 3D or IMRT) with active motion management, OR total skin electrons, OR mixed electron/photon field(s), including imaging guidance, when performed.

Four independent paths into Level 3. Each path is enough on its own. You do not need multiple isocenters and active motion management and mixed fields. One qualifying criterion, actually performed and documented for that fraction, is the standard.

ASTRO Winter 2026 also published final-rule RVUs and national payment examples used for specialty communication at the start of 2026 (77402 RVUs 2.38 / example payment $79.49; 77407 RVUs 9.51 / $317.64; 77412 RVUs 11.72 / $391.46), plus APC movement under OPPS (77412 moving into a higher APC with a much higher example OPPS rate than 2025). Treat those dollars as historical reference from the Winter 2026 article, not as your current locality allowables. Pull your live PFS locality file and OPPS addendum before you price a contract or a denial appeal.

What “active motion management” actually means

This is where most Level 3 denials live.

ASTRO Winter 2026 defines active motion management for 77412 as treatment delivery that includes intrafraction localization and tracking of the target(s) or patient motion to optimize beam delivery (examples given: intrafraction motion, surface guidance). The same article distinguishes setup imaging from intrafraction work: intrafraction motion management uses fiducials or imaging to monitor the target or organs at risk during the breathing cycle (for example during deep inspiration breath hold), to minimize organ motion and allow more accurate delivery to mobile targets and active avoidance of organs at risk.

Three practical implications fall out of that definition:

1. Setup-only surface guidance is not Level 3. Positioning the patient with SGRT before beam-on, then delivering without intrafraction tracking or gating, is not “active motion management” for 77412. It may be excellent clinical practice. It does not, by itself, meet the delivery-code criterion ASTRO published.
2. Vendor does not matter. Behavior does. No particular commercial system is required. Gating, internal surrogate tracking, surface-guided intrafraction monitoring, and other real-time methods can qualify when they actually manage beam delivery during the fraction.
3. “DIBH” as a label is not documentation. Deep inspiration breath hold can be part of an active motion management workflow. Writing “DIBH” in the note without describing monitoring, tolerances, and beam-on / beam-off logic is exactly what utilization managers are rejecting.

What “multiple isocenters” actually means

ASTRO’s early-2026 denial tracking found a second, equally damaging misread: payers treating “multiple isocenters” as “multiple anatomical sites.”

ASTRO Winter 2026 defines the isocenter as the reference point of the treatment machine’s coordinate system. After image verification, if a plan-directed shift is required for additional treatment volumes described in the isodose plan(s), that is a second isocenter.

So:

• Two plan isocenters for volumes in one disease site can be Level 3.
• “Breast and SCLV” as anatomy labels without plan-directed isocenter shifts is not automatically Level 3.
• Counting CTVs, PTVs, or anatomic regions is the wrong proxy. Counting plan isocenters after image verification is the right one.

If your prior-auth packet says “multiple sites” when the medical reason is “multiple isocenters,” you invited the denial ASTRO has been fighting since January.

The denial pattern, in the societies’ own numbers

ASTRO survey (press release dated April 1, 2026)

ASTRO reported that federal officials estimated roughly a 1% specialty payment reduction from the coding changes, while more than two-thirds of surveyed physicians reported declines of 10% or greater, with many clinics anecdotally citing 20-30%. Half of respondents said advanced radiation treatments are frequently delayed or denied, often tied to prior authorization or trouble processing the new codes. ASTRO Board Chair Sameer Keole, MD, FASTRO, called the findings a serious and immediate threat to access.

ASCO (April 7, 2026), citing ASTRO’s national survey

• More than 90% of respondents reported Level 3 interpretation problems
• About 50% of high-complexity claims were flagged or denied
• CMS had projected about a 1% specialty impact; many practices reported revenue drops of 10% or more
• Some Medicaid and commercial systems still had not updated

ASTRO prior-authorization notes (January 14 and January 21, 2026)

ASTRO identified recurring issues with CPT 77412 and CPT 77387-26. For 77412, denials largely stemmed from:

• Payer interpretations limiting the code to multiple anatomical sites rather than multiple isocenters for a single site
• Global application of fraction limits
• Inconsistent coverage for clinically necessary scenarios such as multiple joints or breast treatments using active motion management

EviCore (as relayed by ASTRO) flagged non-specific documentation (“DIBH” alone) and inappropriate use of “urgent” designations as contributors to denials.

Separately, 77387-26 denials were issued on claims that IGRT is bundled with delivery or that the professional component is not separately payable in hospital settings, with inconsistent rationales across freestanding and hospital-based sites. ASTRO also reported working with UnitedHealthcare on an incorrect edit affecting professional image guidance.

September 2026 Oncology Practice Management coverage of the same survey wave underscored the community-practice angle: outdated commercial contract definitions, delayed or denied advanced treatments, and consolidation pressure on independent clinics.

None of that is a coding mystery. It is a definition fight plus a documentation fight plus a contract fight.

The EviCore five-element stack (use it even if EviCore is not your reviewer)

ASTRO published EviCore’s recommended elements for approving 77412 on the active motion management path. Build your template around them whether or not EviCore is on the case. They map cleanly to what medical directors ask for in peer-to-peer reviews:

1. Medical necessity for motion management, why this target or OAR requires intrafraction management for this patient (for example left-sided breast with cardiac sparing goals, thoracic or upper abdominal target with documented respiratory excursion, mobile target near a critical structure).
2. Real-time monitoring or gating system used, name the specific system or technology used for intrafraction motion tracking.
3. Workflow or parameters, how motion is monitored during treatment, including thresholds, tolerance levels, and patient-specific setup.
4. Respiratory monitoring and tracking description, whether monitoring is continuous, how motion data is acquired, and how the therapist is alerted to deviations.
5. Beam-on / beam-off methodology, the criteria used to gate delivery, including what triggers beam-off events.

If your note cannot answer those five questions without hunting through three systems, the prior-auth packet will not answer them either.

77387 is professional only, and payers are still confused

Under the 2026 structure, imaging guidance technical services are packaged into 77402 / 77407 / 77412. ASTRO Winter 2026 states that 77387 will be a professional code only, because the technical component is bundled into the updated delivery codes. ASCO’s April brief says the same: professional IGRT consolidates to 77387; technical IGRT is not separately billed with these delivery codes for the rewritten family.

Operational friction remains:

• Some payers demand modifier 26 on 77387 even though the code is already professional-only in this construct; others reject -26 as redundant.
• Some payers incorrectly treat professional IGRT as bundled into delivery.
• Hospital vs freestanding edits are not consistent.

Practical approach for independent clinics:

• Do not bill 77387-TC with 77402 / 77407 / 77412 for Medicare under the 2026 delivery family.
• For commercial plans, keep a payer matrix: required modifiers, whether professional IGRT is payable on the same date as planning and first fraction, and the appeal citation (ASTRO / ASCO / CPT construct).
• When a payer cites “bundled,” ask whether they mean the technical component (correct) or the professional component (usually incorrect under the 2026 design).

Practical implications and workflow for an independent radiation clinic

1. Rebuild charge capture around the three questions

Before charge entry for each fraction, the record should answer:

1. Is this Level 1, 2, or 3 under the CPT descriptors, not under old IMRT logic?
2. If Level 3, which one criterion applies today: multiple isocenters, active motion management, total skin electrons, or mixed electron/photon fields?
3. Was imaging guidance technical work performed (already bundled) and was professional image guidance separately reportable as 77387?

If question 2 cannot be answered from the treatment record, do not drop 77412 “because the plan was complex.”

2. Split documentation across the episode, on purpose

Level 3 proof is rarely one paragraph. Spread it:

• Consult / planning note: medical necessity for motion management or multi-isocenter strategy; OARs at risk; why setup-only imaging is not enough when motion management is planned.
• Simulation: physician review of motion-management setup; device or system selected; patient-specific tolerances discussed.
• Plan documents: isocenter count and plan-directed shifts; motion-management intent covering the target through the relevant phases of motion.
• Daily treatment record: system used; continuous vs sampled monitoring; tolerance breaches; beam-hold events; therapist alerts; physician authentication timing your medical staff agrees meets local policy.

3. Prior authorization packet = Level 3 packet

For payers that still authorize by delivery complexity:

• Lead with the Level 3 criterion in plain language (“single-isocenter IMRT with continuous surface-guided intrafraction monitoring and automated beam hold outside ±X mm / Y% amplitude”).
• Attach the five-element narrative, not a one-word DIBH flag.
• If the criterion is multiple isocenters, attach the plan page that shows isocenters and the verification/shift workflow, not a list of anatomic sites.
• Avoid “urgent” unless the clinical situation meets the payer’s urgent definition. ASTRO relayed that inappropriate urgent designations contribute to denials.

4. Fix the chargemaster and the mapper

• End-date 77385, 77386, 77014, G6001, G6017 for DOS on/after 1/1/2026.
• Map delivery to 77402 / 77407 / 77412 only.
• For non-excepted off-campus PBDs, enforce PN on 77402 / 77407 / 77412 per MM14361.
• Create a hard stop when someone selects 77412 without a structured reason code (AMM / multi-iso / TSE / mixed fields).

5. Renegotiate commercial fees against the new family

ASCO, ASTRO survey commentary, and September 2026 practice-management reporting all point to the same contract problem: many commercial fee schedules still price legacy G-codes or pre-2015 delivery logic. If 77412 pays like an old simple delivery code, or if 77387 pays a fraction of expected professional IGRT, that is a contracting issue, not a “coder education” issue. Bring utilization mix, denial rates, and locality Medicare comparisons to the table.

6. Track denials like a clinical registry

Minimum fields: payer, DOS, code billed, code paid or authorized, denial reason verbatim, criterion claimed (AMM vs multi-iso vs other), appeal outcome, days delayed to first fraction. Without that log, you cannot tell society advocates or your own CFO whether the problem is documentation, payer policy, or both.

Common failures (and what to do instead)

Failure 1: “IMRT = 77412”

Wrong. Single-isocenter IMRT without active motion management is 77407.

Failure 2: “We used SGRT for setup, so it’s Level 3”

Wrong unless intrafraction tracking actually managed delivery.

Failure 3: “Multiple sites = multiple isocenters”

Wrong. Document plan isocenters and verification shifts.

Failure 4: Checkbox DIBH

Wrong. Use the five-element stack.

Failure 5: Billing 77387-TC with the new delivery codes for Medicare

Wrong under the bundled technical construct described by ASTRO and ASCO.

Failure 6: Leaving G-codes active “just in case”

Wrong for DOS on/after 1/1/2026; MM14361 deletes them for the rewritten family (including the PN pathway for non-excepted off-campus PBDs).

Failure 7: Ignoring Medicaid and smaller commercial plans

ASCO flagged lagging system updates. Test claims early; do not assume Medicare logic equals every payer’s edits.

Failure 8: Treating September 2026 like steady state

CMS-1848-P already tees up Radiation Oncology Treatment Delivery (CPT codes 77402, 77407, and 77412) and proton beam delivery for CY 2027 valuation work, and posts CY 2027 PFS Proposed Rule Proxy Inputs for Radiation Services and Proton Beam Services (Updated 07/21/2026) on the CMS-1848-P downloads page. 2026 claim hygiene and 2027 comment strategy are the same operational problem on two clocks.

The other clock: CMS-1848-P comments due September 14, 2026

CMS issued the CY 2027 Medicare Physician Fee Schedule proposed rule (CMS-1848-P) with a comment period closing September 14, 2026 (Federal Register publication July 16, 2026; display July 14, 2026). Nothing in this section is in effect for 2026 claims. Do not change live edits to match proposals.

From the CMS CY 2027 PFS proposed rule fact sheet and the Federal Register docket:

• Two conversion factors continue. The one-year +2.50% CY 2026 bump under the Working Families Tax Cut legislation expires. CMS proposes a qualifying APM conversion factor of $33.17 (−1.19% from $33.57) and a non-qualifying factor of $32.84 (−1.68% from $33.40).
• CMS proposes to reduce payment when a separately identifiable office/outpatient E/M is furnished by the same physician (or same practice) on the same day as a 0-, 10-, or 90-day global procedure: highest service at 100%, other surgical procedure(s) or E/M visit(s) that day at 50%.
• HCPCS G2211 would become modifiers (placeholder MOD1 at +16% on the associated E/M; placeholder MOD2 at +32% for Shared Savings ACO or LEAD Model participant providers, with assignment/benchmark effects described in the fact sheet). Keep billing G2211 as an add-on through December 31, 2026 unless and until a final rule says otherwise.
• The proposed rule’s valuation agenda explicitly includes Radiation Oncology Treatment Delivery (77402, 77407, 77412) and proton beam codes, and CMS posted radiation/proton proxy inputs in the CMS-1848-P download set.

For freestanding radiation clinics, the comment that matters is not a form letter about “cancer care.” It is a short, evidence-heavy note that attaches:

• Your 2026 utilization mix across 77402 / 77407 / 77412
• Denial and downcoding rates for 77412 and 77387
• Direct practice-expense reality (linac, vault, physics, therapy staffing) against the proposed PE direction
• Concrete examples where Level 3 clinical work was authorized or paid as Level 2

File comments at Regulations.gov on docket CMS-2026-2377 / file code CMS-1848-P. Silence is how proxy assumptions harden into 2027 rates.

Recommendations (action list for the next two weeks)

1. Audit 20 recent 77412 claims (or the last month if volume is low). Score each against one Level 3 criterion and the five-element AMM stack when AMM is the path.
2. Rewrite the daily treatment note template so AMM cases cannot close without system, tolerances, monitoring mode, and beam-hold logic.
3. Retrain schedulers and auth staff to say “multiple isocenters” or “active motion management,” never “multiple sites” or “DIBH only.”
4. Build the 77387 payer matrix and stop automatic -TC.
5. Pull commercial contracts that still reference G6001, G6017 or 77385/77386 and open renegotiation.
6. Download the CMS-1848-P radiation proxy inputs and submit a comment by September 14, 2026.
7. Route stubborn denial patterns to ASTRO’s payer-support contacts when the denial contradicts published descriptors (ASTRO has asked members to send inappropriate denials to its health-policy staff).

Independent practices that do not have spare analytics staff can still do steps 1-4 with a coder, a therapist lead, and a physician champion in one focused week. The comment letter (step 6) can be one page if the attachments are real.

Conclusion

The 2026 radiation delivery rewrite was supposed to modernize payment around how treatment is actually delivered. Eight months in, the clinical descriptors are workable, and the revenue cycle is still tripping on old mental models: IMRT equals complex, DIBH equals a checkbox, multiple sites equals multiple isocenters, and professional IGRT somehow vanished into the linac.

Level 3 pays for documented complexity on the fraction you deliver. Prior authorization follows the same rule. Commercial contracts and Medicaid edits still need active cleanup. And before September 14, CMS is asking whether 2027’s valuation of 77402 / 77407 / 77412 should rest on better evidence than silence.

For clinics that need help tightening Level 3 documentation, prior-auth packets, denial appeals, or specialty chargemaster cleanup across radiation oncology, and related specialty RCM in medical oncology, urology, and behavioral health, ALH Billing Solutions (Woodland / Vancouver, WA) already lives in this workflow daily: coding accuracy, prior authorization support, and denial management for independent practices that cannot afford a 10-30% quiet leak on their highest-complexity days.