Withdrawal Order Calculator: Which Account to Draw From First

Most retirees hold money in three kinds of account, and each is taxed differently on the way out. A withdrawal from a traditional IRA or 401(k) is ordinary income, every dollar of it (IRS Publication 590-B). A sale in a taxable brokerage account is taxed only on the gain, and a long-term gain is taxed at 0%, 15% or 20% depending on taxable income (IRS Topic 409). A qualified withdrawal from a Roth account is not income at all (IRS).

The same $110,000 of spending can therefore put very different amounts on a tax return. Because brackets apply one year at a time, a low bracket left unused in a quiet year cannot be carried forward. The order in which the accounts are drawn decides how much income lands in each year, how much of your Social Security is taxed, how large the required minimum distributions become, and what is left in which account for a surviving spouse or for heirs. Our guide to which accounts to draw from first explains the reasoning in prose. This page puts numbers on it: five orders, run year by year through the same tax calculation.

In every strategy the required minimum distribution comes out first, because the law requires it, and each year's withdrawals are sized so that $110,000 is left after federal tax, any state tax and any Medicare surcharge the income causes. Only the choice of account differs.

1. Conventional: taxable, then traditional, then Roth

The textbook order spends the least sheltered money first. The taxable account loses part of its return to tax every year, so emptying it first leaves more money compounding untaxed. Its weakness is timing: it reports very little income while the taxable account lasts, then a great deal once the traditional account has grown and required distributions begin.

2. Roth first: Roth, then taxable, then traditional

This is the order usually considered worst, shown so that its cost is visible. It spends the only money that will never be taxed again, and leaves the traditional account, which is fully taxable, to grow the longest.

3. Proportional: every account, every year

Each year's withdrawal is split across the three accounts in proportion to their balances. Nothing is optimised, but income is spread evenly across the years, so no bracket is wasted early and none is overfilled late. It is a useful benchmark: a strategy that cannot beat it is not doing much.

4. Bracket-fill: traditional up to a ceiling, then taxable, then Roth

Each year draws from the traditional account until taxable income reaches the top of a bracket you choose (12%, 22% or 24%), or until spending is covered, and only then turns to the taxable account and the Roth. The aim is to take pre-tax money out at a known, low rate while that rate is available. You can set the ceiling at the first Medicare IRMAA threshold instead of a bracket.

5. Bracket-fill with Roth conversions

The same, with one addition. In a year when spending is covered before the ceiling is reached, the room that is left is converted from the traditional account to the Roth, and the tax on the conversion is paid by selling from the taxable account. It is a deliberately simple version of a conversion plan; the Roth conversion calculator sizes and times conversions properly and reports a break-even age.

Illustrative example: the people and figures below are hypothetical, not real clients. Numbers are calculated from the stated assumptions.

The calculator opens on a married couple aged 63 and 61, both retired. They hold $700,000 in a taxable account with a cost basis of $450,000, $1,400,000 in traditional IRAs and $300,000 in Roth IRAs. They spend $110,000 a year after tax. The older spouse will claim $44,000 of Social Security at 70 and the younger $26,000 at 67. Born in 1963, the older spouse has required distributions from 75. Everything is in today's dollars, with a 4% return after inflation, to age 95.

The last column values what is left as heirs would receive it: the Roth and the taxable account in full, since inherited investments take a basis equal to their value at death (IRS Publication 551), and the traditional account less an assumed 22% income tax.

Two things stand out. First, the spread in lifetime tax is wide: the conventional order pays $373,636 and bracket-fill $157,598, a difference of $216,038. Second, the spread in what is left is much narrower: $53,464 between the same two, about 1.5% of the total. Tax paid early is money that stops compounding, and the conventional order's larger traditional balance at 95 is only marked down by 22%. For these inputs bracket-fill with conversions ends furthest ahead, $60,308 above the conventional order. Roth first ends $224,119 behind it.

Year by year: conventional against bracket-fill

The table shows where the difference arises. Under the conventional order the couple sell $110,000 from the taxable account this year. Only the gain, $39,286, is income; after the $32,200 standard deduction their taxable income is $7,086 and the tax is $0. Under bracket-fill they withdraw $120,045 from the IRA instead, all of it income, and pay $10,045, with taxable income of $87,845 still inside the 12% bracket. For this couple the ceiling is never reached: taxable income under bracket-fill peaks at $87,845, below the $100,800 top of the 12% bracket. Strategy 4 is in effect “traditional first” here, and a higher ceiling changes only the conversion strategy.

Bracket-fill pays more tax in the early years: by 70 it has paid $64,636 more than the conventional order. Then required distributions begin, the conventional order's tax rises year after year, and the running total turns in bracket-fill's favour at 81. A household that expects a short retirement, or plans to leave the traditional account to charity, is looking at the top half of this table only.

The years between the last paycheck and the start of Social Security and required distributions are when a household has the most control over its taxable income. For the example couple that is 63 to 69. Under the conventional order they pay $0 of federal tax in those 7 years. That sounds like a success, but the 10% and 12% brackets go almost unused: across the gap, $669,791 of income that could have been taxed at 12% or less was not taken. In 2026 the 12% bracket on a joint return runs to $100,800 of taxable income, on top of the standard deduction (Revenue Procedure 2025-32).

Meanwhile the traditional account grows untouched. By the end of the year they are 74 it holds $2,037,688 under the conventional order, against $843,250 under bracket-fill. Required distributions start at 73, or at 75 for people born in 1960 or later (Internal Revenue Code section 401(a)(9)(C)), and each year's amount is the prior year-end balance divided by an IRS life-expectancy factor (IRS). The first one is $82,833 under the conventional order and $34,278 under bracket-fill. By 95 the conventional order's required distribution has reached $129,622, more than the couple need on top of Social Security, so the surplus is taxed and then reinvested in the taxable account.

Bracket-fill spends the gap differently. It pays $61,441 over the same years, at rates of 10% and 12%, and in exchange reaches 75 with a much smaller traditional account and the taxable account unspent. The RMD calculator and the RMD table show the distribution a given balance will force at each age.

Long-term gains are taxed at 0% while taxable income, gains included, is at or below $98,900 on a joint return in 2026 ($49,450 single), and at 15% above that until a much higher threshold (Revenue Procedure 2025-32). Ordinary income is counted first and gains sit on top of it. That is why the conventional order pays so little early on: with no other income, the couple's $39,286 gain falls entirely inside the 0% band.

It is also why mixing the two sources has a cost. The table holds the same sale and the same gain fixed and adds IRA income underneath it, using the calculator's tax engine for the example couple this year.

Up to a point the gain stays free. Past it, each extra dollar of IRA income pushes a dollar of gain from 0% to 15%, so that dollar costs its own bracket rate plus 15%: at $120,000 of IRA income, $28,186 of the gain is taxed and the sale costs $4,228. The strategies here avoid most of that by not mixing: the conventional order sells while income is low, and bracket-fill leaves the taxable account alone until the traditional draw is done. Above $250,000 of income on a joint return, gains can also bear the 3.8% net investment income tax (IRS Topic 559), which the engine includes. The capital gains tax calculator prices a single sale in detail.

Up to 85% of Social Security benefits can be taxable, and the share depends on other income. The test uses “combined income”: other income plus half of benefits. On a joint return, benefits start to become taxable above $32,000 and reach the 85% rate above $44,000 (IRS Publication 915). Those thresholds are not indexed. Inside the phase-in, a dollar of IRA income adds up to $1.85 of taxable income, so the true rate on that dollar is up to 1.85 times the bracket rate. That is the torpedo.

The table prices the example couple's return at 71, the first full year both benefits are paid ($70,000 together), at rising IRA withdrawals.

In the 12% bracket the rate on the next dollar reaches 22.2%. Once 85% of benefits are taxable the effect stops and the rate falls back to the bracket rate, until the senior deduction begins to phase out above $150,000 of income and adds a little again. In the engine's output the effect shows up at 75: the conventional order's first required distribution makes $59,500 of the couple's benefits taxable, against $35,066 under bracket-fill, which arrives at that age with a smaller traditional balance. The taxable Social Security calculator shows the phase-in for your own benefit.

Medicare Part B and Part D premiums carry a surcharge, IRMAA, when income two years earlier was above a threshold: $218,000 on a joint return and $109,000 for a single filer for 2026 premiums (CMS). Unlike a tax bracket it is a cliff: one dollar over the first threshold costs a couple $2,297 for the year. Because of the 2-year lookback, income from the year you turn 63 already counts.

None of the four pure withdrawal orders reaches the first threshold for the example couple. With conversions, it depends on the ceiling chosen:

Filling the 22% bracket converts $189,242 more than filling the 12% bracket, pays $11,484 of surcharges over 6 years, and ends $91,319 lower. Holding income at the first IRMAA threshold instead converts $739,443 with no surcharge and ends at $3,531,492. For this couple, with heirs assumed to pay 22%, the conversions made at 10% and 12% carry the benefit and the ones made at 22% do not pay for themselves. A higher rate for heirs, or a survivor filing single, changes that. The IRMAA calculator and the IRMAA brackets page list every tier.

A couple files jointly for the year one of them dies; from the next year the survivor generally files as single (IRS Publication 501). The single brackets, standard deduction and IRMAA thresholds are about half the joint ones, while income falls by much less: the smaller Social Security benefit stops, but the required distributions carry on at full size.

Setting the first death at 80 for the example couple shows the shift. Under the conventional order the survivor's federal tax at 81 is $20,915, in the 24% bracket, plus $2,885 of IRMAA; had both lived, the couple would have paid $14,546 in the 22% bracket and no surcharge. Under bracket-fill the survivor pays $10,778 that year. The 22% bracket for a single filer ends at $105,700 of taxable income in 2026.

Lifetime tax here includes IRMAA. The conventional order's bill rises by $141,080 and the conversion strategy's by $46,085, because by 80 it has little traditional money left to force income onto a single return. Its lead over the conventional order widens from $60,308 to $188,797. The model's survivor keeps the larger of the two benefits and the same spending target; a real survivor's benefit and budget can differ, as the method notes explain.

The ranking above belongs to one couple. The table reruns the engine for three other households with the same $2,400,000, the same spending and the same Social Security, changing only where the money sits.

Mostly traditional. With $1,900,000 pre-tax, the conventional order pays $444,414 over the plan and bracket-fill with conversions ends $174,945 ahead of it. This is the household the conventional order serves worst: the taxable account is too small to carry the gap years, and whatever is not drawn early comes out later as large required distributions.

Mostly taxable. With only $500,000 pre-tax, the conventional order ends furthest ahead and bracket-fill ends $131,720 behind the conventional order. Required distributions from a small traditional account do not crowd the brackets, gains on the taxable sales are taxed at 0% or 15%, and what remains in the taxable account passes to heirs with a stepped-up basis. Drawing the traditional account first gives up tax-deferred growth and saves little tax.

Mostly Roth. Here the conventional order ends furthest ahead, and Roth first ends $255,169 behind it. Every order pays little tax; what separates them is how long the Roth is left to grow.

Two assumptions move the ranking as much as the account mix. One is the rate heirs pay on inherited traditional money. For the example couple, the order that ends furthest ahead at each rate is: 0%, the conventional order; 12%, the conventional order; 22%, bracket-fill with conversions; 32%, bracket-fill with conversions. A traditional account bound for charity, or for heirs in a low bracket, is worth more left alone. The other is state tax. At a flat 5% the example couple pay $141,601 of state tax under the conventional order and $102,709 under bracket-fill, and bracket-fill ends furthest ahead. A flat rate is a rough stand-in, since many states exempt Social Security, some exempt pension or IRA income up to a limit, and several have no income tax; the state retirement tax pages give each state's rules.

The engine holds one return constant for the whole plan. Real portfolios do not behave that way, and the order of good and bad years matters most in the first decade of withdrawals; see sequence of returns risk and test the plan in the Monte Carlo retirement calculator. It also treats the three accounts as holding the same investments. Where each asset sits is a separate decision.

Brackets, deductions, IRMAA tiers and the senior deduction are held at this year's rules for every year of the plan (2026 tax brackets). Under current law the senior deduction ends after 2028, so the engine is a little generous to income taken at 65 and later. Before 65, Marketplace premium credits can be worth more than any bracket saving and are not modelled. Qualified charitable distributions, annuities, employer stock and inherited IRAs are left out, and the tax on heirs is a single assumed rate.