news · public-policy · medicare · irmaa · retirement-tax

The IRMAA cliff: how one extra dollar of retirement income can cost you thousands in Medicare premiums

IRMAA isn't a tax — it's a cliff. Cross a Medicare income threshold by a single dollar and your annual premium can jump $1,148 to $1,736 per person, in full, with no phase-in. Here are the official 2026 brackets, the 2-year lookback that makes it stealthy, and the legal ways retirees stay under each tier.

By Mindaugas Laucius · May 27, 2026 · Last reviewed May 27, 2026

Margaret retired in 2024 after thirty-five years as a hospital administrator. Her 2024 modified adjusted gross income was $108,500 — pension, modest Social Security, and a planned IRA withdrawal, sitting just below the first 2026 Medicare income threshold. Then in December 2024, with the stock market up sharply, she sold a chunk of appreciated company shares to help a grandson with a down payment. The capital gain pushed her MAGI to $109,500.

Her 2026 Medicare bill went up by $1,148.40.

Not 1%. $1,148.40 in absolute dollars, every month of 2026, on top of the standard premium. She had crossed the first IRMAA threshold by $499 — and triggered a surcharge that doesn't phase in. It applied in full the moment she crossed.

Welcome to the IRMAA cliff.

This article explains what IRMAA actually is, why it behaves as a cliff and not a curve, exactly where the 2026 cliffs sit, the two-year lookback that makes the whole thing stealthy, and the legal strategies retirees use to stay under each tier. Every number on this page is the 2026 CMS-final figure, sourced from the 2026 Medicare Parts B Premiums and Deductibles fact sheet. Yearfold's own methodology page cites the same source and the calculator uses these brackets directly in every simulation.

Jump to a section: what IRMAA actually is · the 2026 brackets · why it's a cliff · the 2-year lookback trap · the cliff, in one chart · strategies to stay under · how Yearfold models it · common mistakes · FAQ

What IRMAA actually is

IRMAA stands for Income-Related Monthly Adjustment Amount. It is a surcharge that CMS adds to Medicare Part B and Part D premiums for beneficiaries whose modified adjusted gross income exceeds certain thresholds. The Part B surcharge is collected by the Social Security Administration alongside the standard premium (usually deducted from the monthly Social Security benefit). The Part D surcharge is paid directly to your private Part D prescription drug plan, but the amount is set by CMS, not the plan.

The standard 2026 Part B premium is $202.90 per month, per CMS's 2026 fact sheet. The standard 2026 Part D national base beneficiary premium is $38.99 per month, per CMS's 2026 Parts C and D announcement. That works out to a baseline Medicare cost of about $2,903 per person per year before any surcharge. IRMAA can add anywhere from $1,148 to nearly $6,936 on top of that, depending on which tier you land in.

The official rules live in SSA POMS HI 01101.020, which spells out the computation. The mechanic: SSA pulls your modified adjusted gross income from the tax return you filed two years earlier, looks up the IRMAA tier, and applies the surcharge to your current year's premium. MAGI for IRMAA purposes is your adjusted gross income plus tax-exempt interest — most commonly municipal bond interest, which famously isn't taxed federally but still counts for this calculation.

Roughly 7% to 8% of Medicare Part B enrollees pay IRMAA in any given year, per the SSA Trustees Report. That's a minority, but a growing one — the bracket thresholds are indexed to inflation, but income for the top decile of retirees tends to grow faster than the indexing formula, which means more beneficiaries drift into a surcharge tier each year. For the broader context on what Medicare actually costs in 2026 beyond IRMAA — base Part B and Part D premiums, Medigap vs Medicare Advantage, the roughly $370–$520/month a typical 65-year-old retiree pays — see Medicare and IRMAA: the retirement cost most calculators miss.

The 2026 brackets

Every IRMAA bracket discussion online either lists the numbers and moves on, or lists them and then doesn't explain how to read them. Here are the 2026 brackets — based on your 2024 modified adjusted gross income — with each tier's per-person monthly Part B surcharge, Part D surcharge, and the total annual cost (Part B + Part D + IRMAA) each tier produces.

TierMAGI single (2024)MAGI joint (2024)Part B surchargePart D surchargeTotal per person, per year
1≤ $109,000≤ $218,000$0$0~$2,903
2$109,001 – $137,000$218,001 – $274,000$81.20$14.50~$4,051
3$137,001 – $171,000$274,001 – $342,000$202.90$37.50~$5,787
4$171,001 – $205,000$342,001 – $410,000$324.60$60.40~$7,523
5$205,001 – $499,999$410,001 – $749,999$446.30$83.30~$9,258
6≥ $500,000≥ $750,000$487.00$91.00~$9,839

Two things to notice in this table that most explainers don't call out. First, the first cliff adds about $1,148 per year per person. For a married couple where both spouses are enrolled in Medicare and the household's joint MAGI crosses the first tier, that's about $2,297 per year for the household, every year, in surcharges alone. Second, the top tier (MAGI at or above $500,000 single, $750,000 joint) costs about $9,839 per person per year, total, with nearly $6,936 of that being IRMAA. For a high-income couple, the household bill is just under $20,000 per year.

The bracket thresholds are indexed to inflation under a formula CMS applies each November when next year's brackets publish. The indexing is conservative — typically based on the previous year's change in the consumer price index — so households whose income grows faster than inflation will routinely drift into higher tiers even if their inflation-adjusted lifestyle hasn't changed.

Why it's a cliff, not a curve

The standard intuition about tax brackets — that crossing a threshold only taxes the amount over the threshold at the higher rate — is correct for federal income tax. It is wrong for IRMAA.

Walk through the arithmetic for the first cliff at $109,001 single:

  • Single filer with MAGI of $109,000: Tier 1. Part B premium $202.90 per month. Part D base $38.99 per month. Total per-person cost about $2,903 per year. IRMAA surcharge: $0.
  • Single filer with MAGI of $109,001: Tier 2. The full Part B surcharge of $81.20 per month and the full Part D surcharge of $14.50 per month apply. Total per-person cost about $4,051 per year. The extra dollar of income cost $1,148.40 per year in extra Medicare premium.

That is an effective marginal "tax rate" on a single dollar of income of approximately 114,840 percent. There is no other corner of the U.S. tax code where the marginal rate on a single dollar reaches anywhere near this magnitude.

The second cliff at $137,001 single is even steeper: the jump from tier 2 to tier 3 adds another $121.70 per month in Part B surcharge and $23.00 per month in Part D — about $1,736 per year above tier 2 for crossing by a single dollar. The marginal rate on that dollar: roughly 173,640 percent. The cliff at $171,001 adds another ~$1,735 per year for the same one-dollar crossing. So does the cliff at $205,001.

For a married couple both enrolled in Medicare, every cliff is doubled at the household level, because IRMAA is assessed per beneficiary. A couple's joint MAGI crossing the first cliff means both spouses pay the tier 2 surcharge, every month, for the entire year. That's about $2,297 in household IRMAA for crossing the first cliff by one dollar of joint income.

This is the load-bearing fact most people miss when they hear the word "bracket": IRMAA is not a marginal-rate system. It is a step function with extremely sharp risers.

The 2-year lookback trap

This is the part most explainers bury or skip. CMS sets your IRMAA based on the modified adjusted gross income from your tax return filed two years earlier.

  • 2026 Medicare premiums use the MAGI from your 2024 tax return.
  • 2027 Medicare premiums will use the MAGI from your 2025 tax return.
  • 2028 Medicare premiums will use the MAGI from your 2026 tax return.

The rule lives in SSA POMS HI 01101.020. The two-year delay exists because IRMAA is set in late autumn for the following calendar year, and the most recent finalized tax return at that point is two years prior — but the practical effect is that a one-time income event hits you not in the year you incurred it, but two years later, when you may well have forgotten about it.

A Roth conversion in 2024 affects 2026 premiums. A capital gain in 2025 affects 2027 premiums. The sale of a small business in 2023 affected 2025 premiums. An inherited-IRA distribution under the SECURE Act's 10-year rule, if taken in a single year, affects the premium two years later. None of these arrive in the same calendar year as the income event.

This bites hardest for early retirees, whose low-income years before Medicare are the prime Roth-conversion window — exactly where a mistimed conversion can trip IRMAA two years later. If you're weighing whether early retirement is even on the table, run the numbers first.

There is an appeals process — Form SSA-44, "Medicare Income-Related Monthly Adjustment Amount – Life-Changing Event" — but it has a narrow, statutory list of qualifying events. The official form PDF lists exactly eight qualifying life-changing events:

  • Marriage
  • Divorce or annulment
  • Death of your spouse
  • Work stoppage
  • Work reduction
  • Loss of income-producing property (not from sale, gambling, or investment loss)
  • Loss of pension income
  • Employer settlement payment

What is not on the list, and is the single most common reason retirees believe they can appeal IRMAA and then learn they cannot: a one-time spike in MAGI from a planned capital gain, a Roth conversion, a large RMD, or the sale of an investment. None of those qualify as life-changing events under SSA's definition. The income spike is permanent on your tax return, even though your recurring income returns to its normal level the following year.

Form SSA-44 is genuinely useful for someone who retires mid-year, loses a pension, becomes a widow, or otherwise has a permanent change in their income going forward. It is not a way out of a planned, voluntary income event.

The cliff, in one chart

Most charts of IRMAA brackets render the brackets as a smooth curve. That is misleading — IRMAA is fundamentally a step function. Below, the same data as the bracket table above, rendered the way it actually behaves: horizontal across each tier, vertical jumps at each cliff. There are no diagonals. There is no phase-in.

The IRMAA cliff, in one chart.

Total annual Medicare cost (Part B + Part D, per person, single filer) as a function of 2024 MAGI. Each vertical jump is a cliff — cross a threshold by a single dollar and the higher premium applies in full for the entire year. There is no phase-in.

The IRMAA cliff: total annual Medicare cost by MAGIA step chart of total annual Medicare Part B + Part D cost per person, single filer, as a function of MAGI from $80k to $550k. The line stays flat at about $3k until MAGI reaches $109,001, then jumps vertically to $4k, and so on through six tiers. The top tier (MAGI ≥ $500,000) costs about $10k per year. Each vertical jump is the IRMAA cliff — there is no phase-in between tiers.$0$2k$4k$6k$8k$10k$100k$150k$200k$250k$300k$400k$500kMAGI (single filer, 2024 — drives 2026 IRMAA)+$1k/yr+$2k/yr+$2k/yr+$2k/yr+$581/yrTop tier: $10k/yr
View data table
TierMAGI (single, 2024)Part B surcharge / moPart D surcharge / moTotal annual cost
Tier 1≤ $109,500 − $1$0$0$2,903
Tier 2$109,500 − $137,499$81$15$4,051
Tier 3$137,500 − $171,499$203$38$5,787
Tier 4$171,500 − $205,499$325$60$7,523
Tier 5$205,500 − $499,999$446$83$9,258
Tier 6≥ $500,000$487$91$9,839

Source: CMS — 2026 Medicare Parts B Premiums and Deductibles fact sheet · Last reviewed May 27, 2026

The chart is generated from the same constants the Yearfold calculator uses for every simulated year — so the dollars on the chart, the dollars in the methodology page, and the dollars the Monte Carlo applies inside the engine are all the same source of truth. The chart's data table (collapsed by default below the chart) spells out the exact per-tier surcharge that produces each riser.

Strategies to stay under the cliff

A few notes before this section. These are planning concepts that many retirees and fee-only advisors use to manage where MAGI lands in any given year. None of this is personalized advice — every household's tax picture, account mix, and longevity expectation are different. The concepts below are widely discussed in retirement planning practice; whether any of them fits your specific situation is a conversation for a CPA or a fee-only CFP, not a blog post.

Income smoothing across multiple years. Many retirees who are planning large Roth conversions or other discretionary MAGI events spread them across multiple years to keep each year just under a specific IRMAA tier. A single $200,000 Roth conversion done in one year by a married couple with $150,000 of other MAGI lands the household squarely in tier 4. The same $200,000 done as $40,000 each year for five years, with careful monitoring, can keep all five years in tier 1 or tier 2 — saving roughly $9,000 per spouse per year, per year of higher tier avoided. The math compounds when you remember both spouses pay IRMAA at the household tier. The Roth conversion ladder explainer covers when this strategy pays at all, when it doesn't, and the bracket-fill arithmetic.

Qualified Charitable Distributions (QCDs). For retirees who have reached the age where they are taking required minimum distributions from traditional IRAs, a QCD lets you send up to $111,000 per individual in 2026 directly from your IRA to a qualified charity. The amount transferred satisfies the RMD requirement but does not appear in AGI, and therefore does not count toward MAGI for IRMAA purposes. The 2026 limit was set under IRS Notice 2025-67, increased from $108,000 in 2025. Couples can each do up to $111,000 from their own IRAs. There is also a one-time election for a split-interest charitable entity gift of up to $55,000 in 2026.

Tax-loss harvesting. A common practice in taxable brokerage accounts: realize losses on positions trading below their cost basis to offset realized gains. The losses reduce net capital gain (reducing AGI and therefore MAGI). The IRS allows up to $3,000 of net capital loss to offset ordinary income each year, with the remainder carried forward indefinitely. Loss harvesting becomes especially useful in a year you are also realizing a planned gain that would otherwise push you over an IRMAA cliff.

Roth conversion timing relative to age 63. Because of the two-year lookback, the first year a retiree's Roth conversion income flows through to a Medicare premium is the year they turn 65. For someone retiring in their late fifties or early sixties, the "free" years for aggressive Roth conversions are the ones BEFORE age 63 — because MAGI from age 63 and later is what determines premiums in the first Medicare years (ages 65 and 66). Many retirees front-load conversions in the early retirement years for exactly this reason.

HSA contributions, before Medicare enrollment. Workers who are still earning income at age 64 and not yet enrolled in Medicare can contribute to a Health Savings Account, and that contribution reduces AGI (and therefore MAGI). The catch: HSA contributions must stop entirely once you enroll in Medicare. There is no "partial year" deal. For someone delaying Medicare enrollment past 65 because they are still covered by an employer plan, this is a planning lever; for someone already enrolled, it is not.

Asset location. Holding income-generating assets (bonds, REITs, high-dividend funds) inside tax-deferred accounts rather than in a taxable brokerage account means the dividends and interest those assets generate don't count toward AGI in the year they're earned. Many advisors describe this as "location" planning, distinct from "allocation" planning — the asset mix stays the same, but which account each asset lives in changes whether the income shows up in MAGI before retirement-spending withdrawals begin.

The appeal play, when it actually fits. Form SSA-44 works for permanent income changes. Someone who retired mid-2025 and whose 2024 tax return showed full-year salary income may be paying a 2026 IRMAA premium based on a salary they no longer earn. SSA-44 with the "work stoppage" event and a current year's projected lower income can produce an adjustment. Same for a recently widowed spouse whose joint MAGI no longer applies. The form is here; it requires documentation of the qualifying event and a sworn estimate of the current year's MAGI.

How Yearfold models IRMAA

Yearfold's Monte Carlo runs MAGI projections year-by-year across every simulated retirement path. Each simulated year computes the household's MAGI from the prior year's tax return — including projected Social Security taxation, RMDs, Roth conversion income under the user's chosen strategy, and any modeled capital gains — and then looks the result up in the same 2026 IRMAA bracket table this article cites.

The premium the engine applies for each simulated year, for each Medicare-enrolled member of the household, is the standard Part B plus the standard Part D plus the IRMAA surcharge from the tier the prior year's MAGI landed in. That means the success-probability number on the results page already includes IRMAA drag — it's not a separate line item. A plan that drifts into tier 3 in many of the simulated paths produces a meaningfully lower success probability than the same plan that stays in tier 1 or tier 2.

The methodology page cites the same CMS source as this article and lists the bracket thresholds in the Medicare section. When CMS publishes the 2027 brackets in November 2026, the engine, the methodology page, the chart on this post, and any future IRMAA-related content all update from a single source file — lib/taxes/irmaa.ts — so the numbers never drift out of sync.

A future planned feature (not yet shipped at the time of this writing) will surface IRMAA-specific reports: which simulated paths crossed which tiers, in which years, and how the user's chosen income-smoothing strategy moves the distribution. For now, the aggregate effect is already baked into the success-probability output.

Common mistakes retirees make with IRMAA

A scannable list of the patterns that come up repeatedly. Each of these has cost someone real money at some point.

  • Assuming IRMAA is a "tax" and that deductions help. Standard itemized or above-the-line deductions that reduce taxable income but not AGI (the QBI deduction, for example) do not reduce MAGI for IRMAA. Only deductions that reduce AGI — Traditional IRA / 401(k) contributions, HSA contributions, deductible self-employment expenses — actually help.
  • Forgetting the 2-year lookback when planning a Roth conversion at age 63 or 64. Conversions in those years drive Medicare premiums at ages 65 and 66 — the first two years on Medicare. Some retirees converted aggressively right before enrollment and were surprised by the corresponding premium.
  • Believing a one-time capital gain or Roth conversion can be appealed. It cannot. Form SSA-44 only covers the eight qualifying life-changing events listed earlier; voluntary income events are not among them.
  • Thinking only one spouse pays IRMAA in a married household. Both Medicare-enrolled spouses pay the tier-based surcharge. The household's joint MAGI determines the tier; the surcharge applies per beneficiary. A high-MAGI couple in the top tier pays just under $14,000 per year in IRMAA surcharges alone, before any standard premium.
  • Confusing MAGI with AGI when self-calculating. MAGI for IRMAA is AGI plus tax-exempt interest. The tax-exempt municipal bond interest line that doesn't show up in federal taxable income absolutely shows up here, and can push a retiree across a cliff that they thought they were comfortably under.
  • Missing the QCD timing rule. A QCD only counts toward the RMD if the QCD transfer is completed before the RMD has otherwise been satisfied for the year. Doing the regular RMD first and then a separate QCD does not reduce that year's MAGI through the QCD mechanism (though the QCD itself still avoids being added to AGI).
  • Not counting tax-exempt interest in MAGI projections. Many retirement calculators that model "after-tax income" ignore tax-exempt interest because it isn't taxable federally. For IRMAA it counts in full.

Conclusion

IRMAA isn't a tax. It's a cliff with a two-year delay. Most general retirement-calculator products either ignore it entirely or model it as a smooth curve, which misses the load-bearing fact: a single dollar over a threshold costs about $1,148 to $1,736 per person, per year, in surcharges that apply in full from the first day. For high-income households, the bill at the top tier crosses $13,000 per couple per year — every year that joint MAGI stays above $750,000.

The detail matters because the strategies that mitigate IRMAA — income smoothing, QCDs for retirees taking RMDs, the timing of Roth conversions relative to the two-year lookback — are all things retirees would otherwise be doing for general tax efficiency anyway. IRMAA just sharpens the case: an extra $5,000 of voluntary income in the wrong year can cost $6,000 to $7,000 in extra premiums two years later. Knowing where the cliffs sit, and where the household is relative to each one, turns "I'm probably fine" into a number worth running.

Yearfold's Monte Carlo models IRMAA against the 2026 brackets for every simulated year of every path → If you've never seen what IRMAA does to your plan in the unlucky scenarios — the ones where capital gains land in the wrong year and the surcharge follows two years later — that's the place to look. The methodology page cites every primary source the engine uses.

Frequently asked questions

Is IRMAA a tax?

Technically no — IRMAA is an income-related adjustment to your Medicare Part B and Part D premiums, paid through Social Security or directly to your Part D plan. Functionally it behaves like a tax: higher modified adjusted gross income means a higher monthly premium, with no phase-in and no deductions that can reduce it directly.

Can I appeal a one-time capital gain or Roth conversion that triggered IRMAA?

No. Form SSA-44 only accepts appeals based on "life-changing events" that produce a permanent change in income — marriage, divorce, death of a spouse, work stoppage or reduction, loss of pension income, or an employer settlement payment. A one-time capital gain, Roth conversion, large RMD, or inherited-IRA distribution does not qualify, no matter how much it raised your MAGI for that single year.

What income counts toward MAGI for IRMAA?

For IRMAA purposes, MAGI is your adjusted gross income from your tax return plus tax-exempt interest (most commonly municipal bond interest). Capital gains, Roth conversions, RMDs, pension income, Social Security taxable portion, and ordinary investment income all count. Pre-tax 401(k) and Traditional IRA contributions still reduce AGI and therefore reduce MAGI.

Do both spouses pay IRMAA when filing jointly?

Yes — IRMAA is assessed per beneficiary. If both spouses are enrolled in Medicare and the joint MAGI puts the household in, say, tier 3, then both spouses pay the tier-3 surcharge each month. The joint MAGI determines the tier; the surcharge then applies to every enrolled beneficiary in the household.

How long does the 2-year lookback last?

It is a permanent feature of how IRMAA works, not a temporary phase-in. Every year's Medicare premium is based on the MAGI from the tax return filed two calendar years earlier. 2026 premiums use 2024 MAGI; 2027 premiums will use 2025 MAGI; and so on. Planning for a Roth conversion at age 63 or 64 means planning for an IRMAA hit at age 65 or 66.

Sources cited


Last reviewed: May 27, 2026. This article uses CMS-final 2026 Medicare premium and IRMAA bracket figures and 2026 IRS-published QCD limits. Tax and benefit rules are complex and individual situations vary — consult a CPA or fee-only CFP for personalized advice. Yearfold is a financial-education tool. It is not a registered investment adviser and does not provide personalized investment, tax, or legal advice. Results are probabilistic projections based on historical data and stated assumptions; they are not guarantees.

Yearfold is a financial-education tool. It is not a registered investment adviser and does not provide personalized investment, tax, or legal advice. Results are probabilistic projections based on historical data and stated assumptions; they are not guarantees. Methodology

Run the calculator