Retirement Account Withdrawal Order: Taxable, Traditional, or Roth First?

Most retirement planning guides agree on the default withdrawal sequence: spend taxable brokerage accounts first, draw traditional IRA and 401(k) income second, and let your Roth sit untouched as long as possible. The logic holds up — in the abstract. But "correct in general" and "optimal for your balance sheet" diverge substantially once you add Medicare IRMAA thresholds, ACA premium cliffs, or a traditional IRA large enough to generate required minimum distributions bigger than you actually need.

The gap between following the default and running the numbers can exceed $100,000 in lifetime taxes on a mid-size portfolio. Here is when the default is right, when it is not, and the numbers that tell you which situation you are in.

Why the default sequence makes sense

The conventional order — taxable → traditional → Roth — has genuine logic behind it.

Taxable brokerage accounts generate dividends and realized gains every year, creating tax drag whether you spend the money or not. Spending them first eliminates that drag and lets you harvest long-term gains at 0% federal rate if your taxable income stays below $49,450 (single filers, 2026) or $98,900 (married filing jointly). That threshold is achievable in early retirement before Social Security begins, which makes the first several years of taxable spending essentially tax-free on the gain component.

Traditional IRA and 401(k) accounts grow tax-deferred, but every dollar withdrawn is ordinary income. Delaying withdrawals lets the pre-tax balance compound longer. The problem is that the IRS doesn't allow indefinite deferral: required minimum distributions begin at 73 and grow each year. A large traditional balance that compounds untouched for a decade can produce RMDs in the $100,000+ range — taxable as ordinary income whether you need the money or not.

Roth accounts compound tax-free, carry no RMDs, and pass to heirs income-tax-free. Spending Roth last maximizes all three advantages simultaneously.

The default is correct as a starting point. What breaks it are specific portfolio configurations and income thresholds.

The RMD crunch

Consider a hypothetical couple — call them David and Maria — who retire at 66 with $2 million in traditional IRAs, $400,000 in a taxable brokerage, and $200,000 in Roth. Under the default, they spend down the taxable account first over several years, then turn to the traditional IRA.

The problem surfaces at 73. Their $2 million traditional balance, compounded at 6% annually from age 66 to 73, grows to roughly $3 million. The IRS uniform lifetime distribution factor at 73 is approximately 26.5, meaning their first required distribution is about $113,000 — taxable as ordinary income regardless of their actual spending needs.

Add Social Security of $48,000 combined (up to 85% of which is taxable once provisional income clears $44,000 for couples), and their MAGI in that first RMD year exceeds $155,000. They are firmly in the 22% bracket, and the distributions grow larger each subsequent year as the RMD factor shrinks. By age 80, a continued 6% return on the remaining balance keeps pushing the RMD higher.

An alternative plan: during ages 66–72, David and Maria deliberately draw from their traditional IRA each year to fill the 22% bracket — not because they need the money, but to reduce the balance that will be forced out later. Those early withdrawals at 22% prevent larger forced distributions at 24% or 32% in their 80s. The Roth grows untouched throughout. The total lifetime tax bill falls even though they paid taxes "sooner," because the early rate was lower than the rate on the avoided future RMD.

The IRMAA cliff

Medicare's income-related adjustment (IRMAA) operates on a two-year lag: your 2026 Medicare premium is set by your 2024 MAGI. A single large IRA distribution, Roth conversion, or capital gain in the wrong year costs you in premiums two years later — even if your income has since dropped.

The 2026 IRMAA tiers for single filers:

MAGI (single filer, based on 2024 income)	Monthly Part B premium
Under $109,000	$202.90
$109,000 – $137,000	$284.10
$137,000 – $171,000	$405.80
$171,000 – $205,000	$527.50
$205,000 – $500,000	$649.20
Over $500,000	$689.90

Married filing jointly thresholds are double the single-filer amounts.

The cliff structure creates high-stakes math at each boundary. A single retiree projecting $108,500 in MAGI who takes an extra $1,000 from their traditional IRA crosses into Tier 1 and pays $974 more per year in Part B premiums alone — plus Part D surcharges — for a gain of roughly $780 after 22% tax on the distribution. The trade doesn't pencil out.

The reverse is sometimes correct: if your income already lands in the middle of an IRMAA tier, withdrawing an additional $20,000–$30,000 of traditional IRA costs no extra Medicare surcharge until the next threshold. In that situation, the conventional argument against drawing traditional early weakens considerably.

The IRMAA interaction doesn't override the bracket-filling logic — it adds a ceiling. Know your IRMAA tier before finalizing year-end distributions or Roth conversions.

The ACA cliff (before Medicare)

For retirees between 55 and 65 who haven't yet reached Medicare, the ACA premium subsidy cliff replaces IRMAA as the binding income constraint. In 2026, the 400% federal poverty level — above which marketplace premium tax credits phase out sharply — sits at roughly $62,000 for a single person and $84,000 for a couple. Even modest traditional IRA withdrawals stacked on top of investment income can clear that line, eliminating thousands in annual subsidies.

In this pre-Medicare window, the optimal sequence frequently reverses. Take Roth distributions first, supplement with taxable, and limit traditional IRA withdrawals to what fits below the ACA subsidy cliff. That means spending tax-free Roth dollars earlier than the default prescribes — but a $15,000–$25,000 annual premium subsidy at risk is worth more than a few extra years of Roth compounding for most early retirees.

The decision, distilled

Situation	Recommended approach
Large traditional IRA, no IRMAA pressure yet	Default sequence, but convert to Roth in 12%–22% bracket before RMDs begin
Near IRMAA threshold ($109k single / $218k MFJ)	Cap traditional withdrawals or conversions below threshold; draw remainder from taxable or Roth
Under 65 on ACA marketplace	Roth first, then taxable; keep traditional withdrawals below ACA cliff
Traditional balance heading for large RMDs at 73	Draw traditional heavily from 65–72; don't reflexively defer it to preserve Roth
Primarily a Roth holder, small traditional balance	Default applies cleanly; low RMD risk; no deviation usually needed
Estate planning priority	Roth last, always — heirs inherit it income-tax-free and without RMDs

The pro-rata wrinkle

Before executing Roth conversions, one assumption is worth checking: whether you have after-tax (non-deductible) contributions sitting in a traditional IRA. Those dollars carry "basis" — they're not taxable again when withdrawn, since you already paid tax on them when you contributed.

The IRS applies a pro-rata rule: you cannot withdraw or convert only the pre-tax dollars first. Every distribution includes a proportional share of basis across all your traditional, SEP, and SIMPLE IRAs combined — not just the one you happen to be touching. If 20% of your aggregate traditional IRA balance is after-tax basis, then 20% of every conversion is automatically tax-free, regardless of your intentions.

This doesn't change the withdrawal order decision fundamentally, but it does change the tax math on each conversion. Years of non-deductible contributions tracked on Form 8606 can meaningfully reduce what you owe on a given year's distributions. Confirm those numbers before assuming a conversion is fully taxable.

Filling the bracket in practice

The most actionable implementation of the withdrawal sequencing strategy is straightforward bracket arithmetic: each year from retirement through age 72, calculate your ordinary income from fixed sources — pensions, Social Security once started, interest, and dividends. Then calculate how much room remains before the top of your current tax bracket.

In 2026, the 22% bracket for married filing jointly extends to $206,700 in taxable income (meaning roughly $238,900 in gross ordinary income after the $32,200 standard deduction). The 12% bracket tops at $96,950 in taxable income. Fill the gap to either ceiling with traditional IRA distributions or Roth conversions, depending on your time horizon and IRMAA exposure.

Early distributions taxed at 12–22% prevent larger forced distributions taxed at 24–32% after RMDs begin. The total tax paid may be similar in dollar terms over time, but shifting it toward lower brackets materially reduces the lifetime bill — and smooths income in a way that keeps IRMAA thresholds in reach.

The Roth conversion calculator does this bracket math against your actual balances and income. The when-can-I-retire planner models how different withdrawal sequences affect portfolio depletion, Social Security timing, and long-run tax drag simultaneously — including the compounding difference between converting now at 22% versus waiting for an RMD at 32%.

The order determines the bracket

People often treat withdrawal sequencing as an afterthought — something to figure out after the "real" retirement decisions about when to retire, what to spend, and when to claim Social Security. But the withdrawal order determines the tax bracket across a 20–30-year drawdown. Getting it wrong in the early years locks in higher rates for decades. Getting it right compounds the Roth balance, compresses the future RMD obligation, and keeps income below the surcharge thresholds most exposed to cliff risk.

The default is right as a starting point. Whether it is right for your balance sheet depends on three numbers: the size of your traditional IRA relative to your expected annual spending, your projected MAGI versus IRMAA and ACA thresholds, and how many years remain before RMDs begin. With those in hand, the sequence becomes a calculation rather than a convention.

Granary models withdrawal sequencing across your actual accounts, brackets, Roth conversion timing, RMD trajectory, and Medicare exposure — rather than assuming a fixed withdrawal rate from a blended portfolio. The lifetime difference between sequences is typically not a rounding error.

This post is planning education, not tax advice; consult a fee-only CPA or financial planner before implementing any withdrawal sequencing strategy.