Retirement Bridge Calculator
The window, the spending, and the coverage
Your result will appear here
Fill in the fields on the left and this updates as you type.
Know what this estimate is based on
- Jurisdiction
- General mathematical model
- Scope and limitations
- Long-range scenario, not a guarantee. Small changes in returns, inflation, fees, taxes, and withdrawal timing can materially change the result.
- Source links checked
- Jul 30, 2026
Built and regression-tested by Smart Tools Lab. It has not been individually reviewed by a licensed financial, tax, or legal professional.
How to use
- 01
Enter the age you stop working and the age your first pension or Social Security payment starts. The years between them are the bridge, and both ends move the answer more than anything else on the page.
- 02
Enter your annual spending not counting health insurance, then any other income already arriving during those years — rent, a small pension, a spouse's salary. The bridge only has to fund the difference.
- 03
Enter the number of people in the household and the modified AGI you expect in each bridge year. That second figure is realised income — withdrawals, conversions, dividends, gains — not what you spend, and it is what the premium credit is measured against.
- 04
Enter the full-price benchmark silver premium for the household. Full price, before any credit: the page computes the credit itself from the statutory fields below it, which already carry the 2026 figures — Medicare at 65, a $15,650 poverty level plus $5,500 per extra person, and a 9.96% cap.
- 05
Open Advanced options to set the real return the pot earns while it drains (it opens at 0, which prices the bridge as cash) and, if you need to, the cliff percentage. Then read the capital at the top, the credit beside it, and the room left before the cliff.
Formula
Poverty level = $15,650 for one person plus $5,500 for each additional person, so $21,150 for a household of two. Your MAGI as a percentage of that sets the applicable percentage, and the table is six bands rather than one straight line: a flat 2.10% below 133% of poverty, then 3.14% rising to 4.19% across 133–150%, 4.19% to 6.60% across 150–200%, 6.60% to 8.44% across 200–250%, 8.44% to 9.96% across 250–300%, and a flat 9.96% from 300% of poverty up to the cliff. Above 400% no percentage applies and the credit is zero. Your own share = MAGI × the applicable percentage; the credit = the benchmark premium less that share, floored at zero; the net premium is what is left of the benchmark. The pot is then the present value of every bridge year's draw — spending less other income, plus the net premium in each year before the Medicare age — discounted at the real return. At a 0% real return that collapses to the plain sum of the draws, which is the right answer for a bridge held in cash.
Example
Stop at 55, first payment at 67, $62,000 of spending outside health insurance, no other income, a household of two, $48,000 of expected MAGI, a $21,600 full-price benchmark silver premium, and 2% typed into the real-return field under Advanced options. $48,000 is 227% of the $21,150 poverty level for two, so the applicable percentage is 7.59% and your own share of the benchmark is $3,644 a year — the credit covers the other $17,956. Twelve bridge years, ten of them before Medicare at 65, discount to $702,172 of capital, of which $33,388 is health premium. Now move the MAGI field and nothing else. At $84,600 — exactly 400% of poverty — your share is capped at 9.96%, $8,426 for the year, and the pot is $745,987. At $85,000 the credit is not reduced, it is zero: the full $21,600 premium in every pre-Medicare year, and $866,689 of capital. Four hundred dollars of income, $120,702 of capital.
Definitions
- Subsidy cliff
- The 400%-of-poverty line above which the premium tax credit is zero rather than tapered. Removed for 2021–2025 by ARPA and the Inflation Reduction Act, back in force for 2026 coverage.
- Benchmark plan
- The second-lowest-cost silver plan available to your household in your rating area. The credit is calculated against its premium, whatever plan you actually buy.
- Applicable percentage
- The share of your income you are expected to pay toward the benchmark plan — 2.10% to 9.96% for 2026, set by where your MAGI falls against the poverty level. The credit is the benchmark premium less that share.
Good to know
The subsidy cliff is back for 2026, and it is the whole page
From 2021 through 2025, an early retiree buying health coverage on the marketplace had a soft landing: ARPA §9661 and then the Inflation Reduction Act removed the 400%-of-poverty limit on the premium tax credit and capped anyone's own share of a benchmark silver plan at 8.5% of income, however high that income was. Those provisions lapsed on 31 December 2025 and no extension was enacted, so for a 2026 coverage year the original §36B table is back in force. Your expected contribution runs from 2.10% of income at the bottom to 9.96% from 300% of poverty up to the line — and above 400% of poverty the credit is not reduced, it is zero. That single discontinuity is why this page exists, and the arithmetic of it is brutal in a way no smooth projection chart can express. Take the household this page opens with: two people, $62,000 of spending, a $21,600 full-price benchmark premium, twelve bridge years from 55 to 67. At $84,600 of modified AGI — exactly 400% of poverty for two — the applicable percentage caps their own share at 9.96%, which is $8,426 for the year, and the bridge needs $745,987 of capital. At $85,000, four hundred dollars higher, there is no credit at all: the full $21,600 in every pre-Medicare year, and $866,689 of capital. Four hundred dollars of realised income costs $120,702 of capital. No other decision available to an early retiree has that shape. An early retiree is also the household most exposed to it. Premiums are age-rated up to a statutory three-to-one band, so a couple in their late fifties or early sixties faces close to the highest premium in the market with no employer paying most of it. And unlike a wage earner, they usually control their own income to the dollar — which cuts both ways: the cliff is avoidable, but only by someone who knows exactly where it is before December.
Which poverty table governs, and why it is last year's
One detail decides whether every other figure on this page is right, and it is the vintage of the poverty guidelines. A 2026 coverage year is not measured against the January 2026 guidelines. Under 26 CFR 1.36B-1(h) the guidelines that apply are the ones in effect on the first day of the open enrollment period preceding the taxable year, and open enrollment for 2026 coverage began on 1 November 2025 — before the January 2026 table existed. So a 2026 plan is scored against the JANUARY 2025 guidelines, and the January 2026 guidelines will govern a 2027 coverage year. The 48-state figures a 2026 coverage year uses are $15,650 for one person, $21,150 for two, $26,650 for three and $32,150 for four, rising by $5,500 per additional person. Four times those is where the credit ends: $62,600, $84,600, $106,600 and $128,600. Getting the vintage wrong is not a rounding error. The January 2026 table starts at $15,960 for one person with a $5,680 increment, which would put the one-person cliff at $63,840 — so a single filer at $63,000 would be told they were comfortably under the line when in fact they are $400 over it and entitled to nothing at all. That is the same shape of mistake as the $400 in the section above, made by the calculator rather than by the taxpayer. Alaska and Hawaii have their own guidelines, roughly 25% and 15% above the 48-state numbers, and the cliff moves up with them — which at these income levels is worth thousands of dollars of headroom. Both poverty figures on this page are editable fields for exactly that reason: put your own state's numbers in and every figure below, including the cliff, recalculates around them. The same applies a year from now, when the table rolls forward and this page needs the 2026 guidelines rather than the 2025 ones.
How the credit is actually calculated
Three quantities produce the credit, and only one of them is about the plan you buy. The first is the benchmark premium: the second-lowest-cost silver plan available to your household in your rating area, at your ages. That is the plan the credit is calculated against whether or not you buy it — you can take the credit to a bronze plan and pocket the difference, or to a gold plan and pay the excess, but the benchmark sets the size of the subsidy. The second is your income as a share of the poverty level for your household size. The third is the applicable percentage that share implies, and its shape is worth seeing properly, because a single straight line from 2.10% to 9.96% is wrong across most of the range. It is flat at 2.10% below 133% of poverty, jumps to 3.14% and rises to 4.19% across 133–150%, runs 4.19% to 6.60% across 150–200%, 6.60% to 8.44% across 200–250%, 8.44% to 9.96% across 250–300%, and then sits flat at 9.96% from 300% of poverty all the way to the cliff. The credit is the benchmark premium less your income times that percentage, floored at zero. Two bands change the advice rather than the number. Between 100% and 250% of poverty, silver plans — and only silver plans — carry cost-sharing reductions that lift a normal 70% actuarial value to roughly 94%, 87% or 73% depending on the band, so in that range the cheapest bronze plan is usually the more expensive choice. And below 100% of poverty there is no premium tax credit at all: in a state that expanded Medicaid you would be on Medicaid, and in one that did not you fall into the coverage gap with neither, which is why a household that controls its own realised income sometimes deliberately realises a little more. Finally, the credit is normally advanced monthly against an estimate and reconciled on the return. Below 400% of poverty there are caps on repaying an over-advanced credit; above 400% there are none, so a December capital gain becomes an April bill for the whole year's subsidy.
Where the money lives decides what the coverage costs
A bridge needs capital you can actually reach before 59½, which is a different constraint from having it, and the route you use also sets the income that sets your premium. There are four ordinary routes. A taxable brokerage account is reachable at any age and generates MAGI only on its dividends and on the GAIN portion of whatever you sell — sell $60,000 of a holding with a $45,000 basis and only $15,000 is income, which is why a taxable account is the most subsidy-friendly pot an early retiree can own. Roth contributions already made come out at any time, at any age, with no clock, no tax and no MAGI at all: the cleanest bridge dollars in existence. A 72(t) series of substantially equal periodic payments unlocks an IRA before 59½ but is fully ordinary income and locks you into a fixed schedule for the later of five years or age 59½. And the rule at IRC 72(t)(2)(A)(v) lets you take from the plan of the employer you separated from in or after the calendar year you turned 55, penalty-free — but it applies to employer plans only, so rolling that balance into an IRA destroys the route permanently, and there is no way to undo it. A 401(k) balance on none of those routes is not bridge capital, however large it is. Because each route has a different MAGI signature, the account you draw from is a health-insurance decision as much as a tax one: $70,000 of spending funded from a traditional 401(k) is $70,000 of income and may put a household over the cliff, while the same spending funded from basis and Roth contributions might be $10,000. Two dates close the window. Medicare at 65 ends the marketplace question and replaces it with Part B, Part D and a supplement — usually smaller, but with a surcharge set by the income you realised TWO years earlier, so the conversions and gains you take at 63 decide your premium at 65. And COBRA can cover the first eighteen months at full unsubsidised cost, which is sometimes the right choice for a mid-year exit, and sometimes far worse than a subsidised marketplace plan.
Frequently asked questions
What changed for 2026?
The subsidy cliff came back. From 2021 through 2025, ARPA §9661 and then the Inflation Reduction Act removed the 400%-of-poverty limit and capped anyone's benchmark premium at 8.5% of income however high that income was. That provision lapsed on 31 December 2025 with no extension enacted, so the original §36B table is in force again: your own share runs from 2.10% to 9.96% of income up to 400% of poverty, and above that line the credit is zero. Not tapered, not reduced — zero. For an early retiree in their late fifties, whose unsubsidised premium is the largest single line in the budget, that is the difference between two very different retirements.
Where exactly is the cliff for my household?
Four times the poverty level for your household size. On the guidelines a 2026 coverage year uses, that is $62,600 for one person, $84,600 for two, $106,600 for three and $128,600 for four. The page computes it from the two poverty fields, so if you are in Alaska or Hawaii — which have their own, higher guidelines, roughly 25% and 15% above the 48-state figures — replace both fields with your state's numbers and the cliff moves up with them.
Why does the page use the 2025 poverty guidelines for a 2026 plan?
Because that is the table the rule points at. Under 26 CFR 1.36B-1(h) the guidelines that govern a coverage year are the ones in effect on the first day of the open enrollment period preceding it, and open enrollment for 2026 coverage began on 1 November 2025 — before the January 2026 guidelines existed. Using the newer table would move the one-person cliff from $62,600 to $63,840 and tell a single filer at $63,000 they were safely under it when they were $400 over and entitled to nothing. The January 2026 figures govern a 2027 coverage year.
What counts as income for the credit?
Modified AGI: your adjusted gross income plus tax-exempt interest, plus any non-taxable Social Security, plus excluded foreign earned income. There is no asset test at all — a household with three million dollars invested and $40,000 of realised income qualifies on the same terms as anyone else. What does count is every dollar you cause to be realised: a traditional 401(k) or IRA withdrawal counts in full, a Roth conversion counts in full, capital gains and dividends count, and interest counts. Withdrawing Roth contributions you already made produces no MAGI at all, and selling a taxable holding produces MAGI only on its gain, not on the whole proceeds.
Where is bridge money allowed to live?
Somewhere you can actually reach before 59½, which is a separate question from having it. The ordinary routes are a taxable brokerage account, Roth contributions already made, a 72(t) series of substantially equal payments, or the rule that lets you take from the plan of the employer you separated from in or after the year you turned 55. A 401(k) balance on none of those routes is not bridge capital however large it is — and rolling that balance to an IRA destroys the age-55 route permanently, because 72(t)(3)(A) expressly denies it to IRAs. The choice between the routes also sets your realised income, which sets the premium above, so it is one decision rather than two.
What if my income lands under 100% of the poverty level?
That is a hazard rather than a triumph. There is no premium tax credit at all below 100% of poverty: in a state that expanded Medicaid you would be on Medicaid instead, and in one that did not you fall into the coverage gap with neither. A household that controls its own realised income can deliberately realise a little more — a small Roth conversion, a harvested gain — to land above the line and buy a subsidised marketplace plan. Between 100% and 250% there is a second reason to be careful which plan you buy: silver plans in that band carry cost-sharing reductions that no other metal level gets.