Primary source: CMS fact sheet, "Fiscal Year 2026 Improper Payments," November 16, 2026. CMS says the figures are in the fiscal year 2026 HHS Agency Financial Report, at hhs.gov/afr. Trade pickup: Becker's Payer Issues, Austin Garza, November 17, 2026.

CMS's FY 2026 Improper Payments Sum to About $112.9 Billion. The Work Is the Record.

On November 16, 2026, CMS published its fiscal year 2026 improper-payment estimates. Medicaid is the largest line on the page: 7.08 percent, or $46.84 billion. Original Medicare is 7.32 percent, or $33.96 billion. Add Medicare Part C, Part D, CHIP, and advance premium tax credits on the federally facilitated exchange, and the six figures sum to about $112.9 billion. The same six a year earlier sum to about $96.1 billion. CMS did not print either total. If you run revenue cycle, do not book this as a fraud statistic. CMS says the rates do not measure fraud. The work that reaches a billing office is a record that was missing, or that did not support the payment.

About $112.9 billion

The sum of the six estimates in the November 16 fact sheet. Original Medicare $33.96 billion, Part C $23.86 billion, Part D $5.16 billion, Medicaid $46.84 billion, and CHIP $2.26 billion add to $112.08 billion. The advance premium tax credit estimate is $830.37 million. Together that is $112.91037 billion, which rounds to about $112.9 billion. The fiscal year 2025 lines on the same page ($28.83 billion, $23.67 billion, $4.23 billion, $37.39 billion, $1.37 billion, and $657.46 million) sum to about $96.1 billion. Becker's reports both totals.

What the November 16 fact sheet reports

The fact sheet is the document. CMS defines an improper payment as a payment that does not meet program requirements. That includes overpayments, underpayments, and payments where there is not enough information to tell whether the requirements were met. CMS says the rates do not measure fraud. Some improper payments may involve fraud or abuse. Others are missing documentation or an administrative error. The example CMS gives is a state or a contractor missing a required step that, if completed, would have shown the payment was proper.

ProgramFY 2026FY 2025, as restated on this pageWhat CMS says drove it
Original Medicare7.32%, or $33.96 billion6.55%, or $28.83 billionTenth consecutive year below the 10 percent compliance threshold. The fact sheet does not print a cause mix for this line.
Medicare Part C5.68%, or $23.86 billion6.09%, or $23.67 billionMost Part C improper payments: the Medicare Advantage organization's supporting documentation failed to substantiate beneficiary diagnosis data submitted for payment. The rate fell. The dollars rose. The fact sheet does not explain that gap.
Medicare Part D3.85%, or $5.16 billion4%, or $4.23 billionMost Part D improper payments: plan sponsors failed to provide sufficient documentation to support the prescription data used for payment. The rate fell. The dollars rose. No explanation of that gap is on the page.
Medicaid7.08%, or $46.84 billion6.12%, or $37.39 billionAbout 79.41 percent of the improper payments were insufficient documentation. The estimate is three review cycles of about 17 states each, from reviews in 2024, 2025, and 2026. It is not a single-year census of every state.
CHIP9.81%, or $2.26 billion7.05%, or $1.37 billionAbout 55.93 percent from insufficient documentation. Composed of reviews in 2024, 2025, and 2026. The tenth-year sentence under the 10 percent line is printed only for Original Medicare.
Advance premium tax credits, federally facilitated exchange, benefit year 20240.84%, or $830.37 millionBenefit year 2023: 0.89%, or $657.46 millionPrimary cause: overpayments from automated errors, 59.24 percent of improper payments, mostly duplicate members and duplicate use of a Social Security number. Secondary cause: manual errors examining documents a consumer submitted. The rate fell. The dollars rose.

CMS attributes part of the Medicaid and CHIP increase to the unwinding of COVID-19 public health emergency flexibilities. Eligibility redeterminations and provider revalidation requirements resumed in April 2023. That sentence covers both programs. It does not assign a dollar share to the unwinding.

Read the "what you need to know" list under the six lines. CMS says most improper payments happened where a reviewer could not tell whether a payment was proper because documentation was insufficient, and it names the parties who did not supply it: a state, a provider, a Medicare Advantage organization, a Part D sponsor, or a Marketplace insurer. The causes CMS lists are no documentation; insufficient documentation; documentation that does not substantiate the payment; a payment to the right recipient for the right amount where the process did not comply with the statute or the regulation; and, for Medicaid and CHIP, no record that a required eligibility factor was verified, income given as the example. A proper payment, in the same list, is one with sufficient documentation. CMS's two examples are an eligibility verification the agency or the state kept and applied, and enough documentation to support the program's payment rules.

The Payment Integrity Information Act of 2019, Public Law 116-117, is the definition CMS prints for a significant improper payment: greater than $10 million and over 1.5 percent of all payments under that program, or greater than $100 million. Each dollar figure on the fact sheet is above $100 million. The 10 percent line is a different test. CMS says Original Medicare has been under that compliance threshold for a tenth straight year, and the footnote says the threshold was originally published under the Improper Payments Elimination and Recovery Act of 2010 and adopted by the 2019 law.

Original Medicare and Medicaid/CHIP measure claims from before the current fraud tools

The fact sheet is specific about the window, and the window is not every program. Fiscal year 2026 Original Medicare and Medicaid and CHIP measurements are based on claims submitted from July 1, 2024, through June 30, 2025. CMS says that period is before many of the fraud initiatives it describes as current. The page does not say Part C, Part D, or the advance premium tax credit measurement uses those same dates.

The initiatives CMS lists, as coming after that claims window, are Medicare enrollment moratoria for high-risk provider types, and CMS names home health agencies and hospices in that group; large-scale enrollment revocations and payment suspensions aimed at suspicious billing patterns; greater use of advanced analytics to identify high-risk Medicaid providers and to coordinate investigations with states and HHS's Office of Inspector General; and new data dashboards that give states what CMS calls a Super Intelligence-powered tool to track claims and catch Medicaid outliers before losses occur. CMS says the effect of these actions, and of its other fraud work, will begin to show in the improper-payment data for fiscal year 2027, with more substantial effects expected in fiscal year 2028.

In the same section CMS describes ClaimsCore. It calls ClaimsCore a best-in-class commercial claims-processing effort that will re-platform Original Medicare claims processing onto a secure, configurable, modern foundation. CMS estimates the work will save millions. It says that estimate does not include additional billions that may be saved through improved program integrity. ClaimsCore, CMS says, is designed to support modern payment-integrity controls, faster and more transparent policy execution, and advanced capabilities such as what it calls Super Intelligence, while improving operational efficiency and care delivery over time. The fact sheet does not give a timeline, a vendor, or a contract value.

What this is not

It is not a fraud rate. CMS says so. It is not a CMS-printed total of $112.9 billion. That figure is the sum of the six program estimates, and Becker's reports the same sum. It is not an overpayment bill you can reserve against. The definition includes underpayments, and the fact sheet does not split any program into overpayments and underpayments. It is not a provider-type breakout. Home health, hospice, inpatient, and physician are not separated on the page. It is not an announcement of new medical-review, Targeted Probe and Educate, or Recovery Audit Contractor audits. It is not the CRUSH proposed rule. That rule, also dated November 16, cites fiscal year 2025 Part C figures and prices a RADV extrapolation. This fact sheet is the fiscal year 2026 measurement. ClaimsCore is not a go-live date.

RevCycleAI analysis: the record is what gets recouped

This is RevCycleAI's read for people who pull charts, code risk adjustment, and post Medicaid eligibility denials. The dominant cause CMS names for Medicaid, CHIP, Medicare Advantage, and Part D is documentation. That is a clinical-documentation, coding, and records-request problem. It shows up in a provider shop as an additional documentation request, a medical-record pull, an audit, and a recoupment. It is not, on this page, a fraud case CMS filed against a named provider.

Do not collapse every missing page onto the hospital. CMS's own summary says the insufficient documentation came from a state, a provider, a Medicare Advantage organization, a Part D sponsor, or a Marketplace insurer. The administrative example is a state or a contractor skipping a step that would have made the payment proper. Some of these dollars never leave a provider's account. The ones that do are the ones where the plan or the reviewer asks the provider for the chart and the chart does not close the question.

Original Medicare rose, and it is still under 10 percent. The rate moved from 6.55 percent and $28.83 billion to 7.32 percent and $33.96 billion. CMS does not say why, and it does not print a cause mix. A higher fee-for-service improper-payment rate makes more medical review, and more Targeted Probe and Educate or Recovery Audit Contractor-style scrutiny, plausible. Plausible is the word. This fact sheet does not announce those audits. The claims in the rate were submitted by June 30, 2025, before the enrollment moratoria and the revocation wave CMS says came later. Those tools decide who can bill. They are not a published schedule of chart audits.

Medicaid's documentation share is an eligibility file as well as a chart. About 79.41 percent of Medicaid improper payments were insufficient documentation, inside three review cycles of about 17 states each. CMS says part of the Medicaid and CHIP increase is the unwinding: redeterminations and provider revalidation resumed in April 2023. The eligibility error CMS names is specific. There is no record that a required factor, such as income, was verified. That record is the state's. The revenue-cycle consequence is a coverage determination that cannot be defended, a recoupment, or bad debt when the patient looked eligible and the file does not show the check. Work-requirement verification widens the same file. CMS's optional hold-harmless through June 30, 2027, covered separately in the November 13 work-requirement notice, still has states that take it running the check from January 1. This fact sheet does not mention work requirements. The link is the eligibility record, which both documents turn on.

Part C is diagnosis support, which is the chart in the separate CRUSH proposal. Most Part C improper payments, CMS says, are cases where the Medicare Advantage organization's documentation did not substantiate the diagnosis data submitted for payment. That is risk-adjustment coding. The same day's CRUSH proposal is a different document. It cites the fiscal year 2025 Part C measurement, which the proposed rule describes as over $23 billion and which this fact sheet prints as 6.09 percent, or $23.67 billion. The proposed rule says that measurement is not a RADV audit and does not recover money. This fact sheet's Part C line is the next measurement: 5.68 percent, or $23.86 billion. It does not split overpayments from underpayments, and it does not mention RADV. Both documents are about a diagnosis the record does not support. This page estimates the improper payment. CRUSH proposes extrapolation of audit findings if that rule is finalized. They are not the same dollar.

Part D is the prescription file. Plan sponsors, CMS says, failed to provide sufficient documentation for the prescription data used to pay the claim. When that record sits with a pharmacy or a prescriber, the request comes back as documentation. The fact sheet does not say how often that happens, and it does not name a drug class or a provider type.

ClaimsCore is a signal that Original Medicare's edits will move, on a date the page does not give. A re-platform of fee-for-service claims processing, described as configurable and tied to payment-integrity controls and faster policy execution, is a new place for the edit to live. CMS estimates millions in savings from the work itself and says additional billions may be saved through program integrity, which it does not fold into the millions. There is no vendor, no contract value, and no quarter. Do not put a go-live on a project plan off this sentence.

Advance premium tax credits do not fit the documentation story. The primary cause CMS names is overpayments from automated errors, 59.24 percent, mostly a duplicate member or a duplicate Social Security number. The secondary cause is a manual miss while examining documents a consumer submitted. That is an exchange payment error. It is not a hospital claim, and the general sentence about insufficient documentation should not be used to overwrite the cause CMS printed for this program.

The RCAI signal

About $112.9 billion is the sum of six CMS estimates, not a total CMS printed, and CMS says the rates do not measure fraud. Medicaid's $46.84 billion is the largest line, and about 79.41 percent of Medicaid improper payments were insufficient documentation. Part C's finding is diagnosis support, which is the chart behind the separate CRUSH extrapolation proposal. Original Medicare rose to 7.32 percent. More medical-review scrutiny is plausible. This fact sheet does not announce it, and it does not split overpayments from underpayments.

RCAI take

Staff the record request, and do not reserve the whole dollar figure as if CMS had called it fraud or called it all an overpayment. The open items are the overpayment-versus-underpayment split the fact sheet does not print, the provider types it does not name, and the ClaimsCore date, vendor, and contract value it does not give. The Original Medicare and Medicaid and CHIP claims in these rates were submitted from July 1, 2024, through June 30, 2025. CMS says the fraud tools it lists will start to show in the fiscal year 2027 data, and more so in fiscal year 2028.

Sources: CMS, "Fiscal Year 2026 Improper Payments," fact sheet, November 16, 2026 (page opened). The definition of an improper payment, including overpayments, underpayments, and payments with insufficient information; the statement that rates do not measure fraud; the state-or-contractor administrative-step example; every rate and dollar in the table above; the tenth consecutive year below the 10 percent threshold for Original Medicare; the Part C diagnosis-documentation finding; the Part D prescription-documentation finding; Medicaid's about 79.41 percent and the three cycles of approximately 17 states from reviews in 2024, 2025, and 2026; CHIP's about 55.93 percent and reviews in 2024, 2025, and 2026; the partial attribution of the Medicaid and CHIP increase to unwinding COVID-19 public health emergency flexibilities, including eligibility redeterminations and provider revalidation that resumed in April 2023; the advance premium tax credit figures for benefit year 2024 and benefit year 2023, the 59.24 percent automated-error share, duplicate member and duplicate Social Security number errors, and the secondary manual-error cause; the July 1, 2024 through June 30, 2025 claims window for the fiscal year 2026 Original Medicare and Medicaid and CHIP measurements; the four fraud initiatives as named, including home health agencies and hospices, enrollment revocations and payment suspensions, analytics with states and HHS-OIG, and state data dashboards described as a Super Intelligence-powered tool; the expectation that effects begin in fiscal year 2027 data and are more substantial in fiscal year 2028; the ClaimsCore description, the estimate of millions in savings, and the additional billions that may be saved through program integrity; the Payment Integrity Information Act of 2019 (Public Law 116-117) significant-improper-payment definition; the cause list and the two proper-payment examples; the statement that most improper payments occurred where insufficient documentation came from a state, a provider, a Medicare Advantage organization, a Part D sponsor, or a Marketplace insurer; and footnote 1 on the Improper Payments Elimination and Recovery Act of 2010 are the fact sheet's. The fact sheet points to the fiscal year 2026 HHS Agency Financial Report at www.hhs.gov/afr and to CMS's improper-payment measurement programs page. This article does not pull additional figures from either page. No provider-type breakdown, no overpayment-versus-underpayment split, and no ClaimsCore timeline, vendor, or contract value appear in the fact sheet, and none is added here. The $112.08 billion subtotal, the $112.91037 billion six-program sum, the rounding to about $112.9 billion, and the prior-year sum of about $96.1 billion are arithmetic on the fact sheet's six lines, labeled as such. Becker's Payer Issues, Austin Garza, "CMS flags $112B+ in improper payments: 9 things to know," November 17, 2026 (page opened). Becker's reports an estimated $112.9 billion in fiscal 2026, up from about $96.1 billion a year earlier, from estimates CMS released November 16. The program rates Becker's repeats match the fact sheet. No quotation is used. The fact sheet contains no quotations, and none is added. The November 16 CRUSH article is a different announcement and is cited only for the fiscal year 2025 Part C measurement and the proposed RADV extrapolation. The Medicaid work-requirement article is a different announcement. The RevCycleAI analysis section is our read of the fact sheet.