Withdrawal Sequencing — The Conventional Order Quietly Builds a Tax Bomb

01 The three tax buckets

Almost every retirement plan splits into three buckets, and the tax treatment of each is the whole game:

• Taxable — brokerage and savings. You already paid tax on the contributions; only the gains are taxed, often at favorable long-term capital-gains rates (including a 0% bracket in low-income years).
• Tax-deferred — traditional 401(k) and IRA. Contributions went in pre-tax, so every dollar out is taxed as ordinary income — and the IRS eventually forces withdrawals via RMDs at 73.
• Tax-free — Roth IRA and Roth 401(k). Contributions were after-tax, so qualified withdrawals are tax-free, and a Roth IRA has no required distributions during your lifetime.

Sequencing is simply deciding, each year, how much of your spending need to draw from each bucket — and the answer changes the tax you pay now and the tax you're setting up for later.

The conventional order is taxable first, then tax-deferred, then Roth — and as a default it is defensible.

The logic is straightforward. Taxable accounts hold assets with a cost basis, so only the gain is taxed and often at preferential long-term rates. Tax-deferred withdrawals are ordinary income. Roth withdrawals are tax-free and carry no required distributions, so leaving them until last maximises the tax-free growth.

Where the default fails is that it ignores brackets. Spending only taxable money through your sixties can leave you in a very low bracket for a decade while a large traditional balance keeps compounding — and then required minimum distributions arrive at 73 and push you into a higher bracket for the rest of your life, taking Social Security taxation and Medicare surcharges with them.

Put differently, the strict order optimises each year in isolation and can produce a poor lifetime result. The years of unusually low income between retiring and starting RMDs are an asset, and spending them at a 10% effective rate wastes bracket space that will be charged at 24% later.

That is why the better approach replaces a fixed order with a yearly decision, which the rest of this works through.

Source: IRS — Roth IRAs

02 The conventional order and why it exists

The order you'll read everywhere is to spend like this: taxable first, tax-deferred second, Roth last. The logic is genuinely sound as a starting point:

1. Taxable first — let the tax-sheltered accounts keep compounding, and pay only capital-gains rates as you spend.
2. Tax-deferred second — you'll be forced to draw it eventually via RMDs, so deferring the ordinary-income tax has value.
3. Roth last — it grows tax-free, never forces a distribution, and is the best dollar to leave to heirs.

It's simple, it's defensible, and for someone with modest balances it's often close to optimal. The problem shows up when the tax-deferred bucket is large: deferring it doesn't make the tax disappear, it just postpones — and concentrates — it.

The taxable-then-deferred-then-Roth order is repeated so consistently that it is treated as settled, and following it strictly is how people end up in a higher bracket at 73 than they were at 63. It optimises each year in isolation while leaving a decade of low brackets unused, and required minimum distributions then arrive at a rate those brackets would have absorbed cheaply.

Source: IRS Publication 590-B

03 Worked example: your pull this year

Here's the single most useful question for one year: how much can I pull from my traditional 401(k)/IRA and still stay inside a chosen bracket? The 2025 standard deduction shelters the first slice of income ($15,000 single, $30,000 married filing jointly), so your taxable income is your tax-deferred withdrawal minus that deduction. Enter your filing status's bracket ceiling and the deduction below to see the room you have.

With a $60,000 need, a $48,475 bracket top (the 2025 top of the 12% bracket for a single filer), and the $30,000 MFJ deduction, the model says fill the entire need from tax-deferred — every dollar comes out at 12% or less. Drop the bracket top and the model starts routing the overflow to taxable or Roth so you don't spill into the next bracket. The point isn't the exact figure; it's that "taxable first" would have left this whole 12% room unused.

Roth is routinely described as the account to spend last, and that is not quite right either. It is the account to spend selectively: because a Roth withdrawal is not income, it is the one source that can top up a year already close to an IRMAA threshold or the Social Security taxation phase-in without pushing you across it. Saving it strictly for the end gives up its most valuable use.

On the defaults above, the worked example shows: Your whole $60,000 need fits inside this bracket from tax-deferred — pulling it now beats letting it compound into a larger RMD.

04 The RMD tax bomb at 73

This is the failure mode the conventional order walks straight into. If you leave the traditional 401(k)/IRA untouched while it keeps compounding, by 73 it can be far larger than when you retired — and the IRS then requires annual distributions, taxed as ordinary income, whether you need the cash or not. Required minimum distributions start at 73 under current law, calculated as your prior-year-end balance divided by an IRS life-expectancy factor.

The early withdrawals you take in your 60s do double duty: they fund your spending at a low rate and shrink the balance that drives future RMDs. That's the lever the rigid order never pulls.

Source: IRS — RMD FAQs

05 Bracket-aware blending and conversions

The tax-smart approach replaces a fixed sequence with a yearly decision: blend the buckets to fill your low brackets deliberately.

In a low-income year you have two ways to use that room. Spend tax-deferred money you would otherwise have left alone, which meets this year's needs at a low rate. Or convert to Roth up to the top of a chosen bracket, which does not fund spending but permanently removes future required distributions on the amount converted. Both shrink the future RMD; the conversion also builds the Roth balance you will spend last.

Both use the same bracket space and you generally cannot do both to the full extent, so the choice is between funding today cheaply and reducing tomorrow's forced income. Which wins depends on how large the traditional balance is relative to what the low-bracket years can absorb.

The method that works is arithmetic rather than judgement: calculate income before any discretionary withdrawal, find the distance to the top of the target bracket, and take exactly that much. Recalculate each year, because Social Security starting, a spouse's death changing the filing status, or a large capital gain all move the answer.

None of this means draining the tax-deferred bucket recklessly. It means using the brackets you're given each year instead of saving them all for a tax authority that will hand them back as a bill at 73.

Source: IRS — Roth IRAs

06 Where IRMAA and Social Security taxation bite

Sequencing isn't only about income-tax brackets — three thresholds turn a clean plan messy if you ignore them, and the source of each withdrawn dollar is exactly what moves you across them.

• Social Security taxation — as your other income rises, up to 85% of your Social Security benefit becomes taxable. A large tax-deferred withdrawal can drag more of an otherwise-untaxed benefit into the tax base.
• IRMAA — high income raises your Medicare Part B and D premiums roughly two years later, in steps. A one-off big RMD or conversion can quietly bump you a tier.
• The 0% capital-gains bracket — long-term gains are taxed at 0% while taxable income stays under about $48,350 single / $96,700 MFJ for 2025, which makes spending or harvesting from the taxable bucket especially cheap in low-income years.

Social Security taxation is the first. Depending on combined income, up to 85% of your benefit becomes taxable, and the phase-in creates effective marginal rates well above the nominal bracket — a stretch where an extra dollar of IRA withdrawal can be taxed at an effective 40.7% while sitting in the 22% bracket.

IRMAA is the second, and it is a cliff rather than a taper: one dollar over a threshold raises Medicare premiums for a whole year, for both spouses, on Part B and Part D, based on income from two years earlier.

The long-term capital gains rate is the third. Below a threshold it is 0%; above it, 15%. A traditional IRA withdrawal that pushes ordinary income up does not itself get taxed at capital gains rates, but it displaces the room that would have let gains be realised at zero.

This is where Roth earns its "spend last" reputation in reverse: a small Roth withdrawal is the one source that tops up a high-income year without nudging Social Security taxation, IRMAA, or the capital-gains threshold.

The buckets are not just tax rates. They are levers for staying on the right side of thresholds that cost far more than the rate difference between them.

Source: IRS Tax Topic 409 — Capital Gains and Losses

07 What each bucket does to your thresholds

The buckets differ in tax rate and, more importantly, in what they do to the thresholds that surround it.

Read the last two columns rather than the second. Two sources move nothing at all, which is exactly what makes them valuable for topping up a year that is already near a threshold.

Source: IRS — Retirement plans