The Right Withdrawal Order Isn't a Checklist. It's Bracket Math, Redone Every Year.

What the order leaves out

"Spend taxable first" treats a taxable brokerage account as one thing. It's a pile of cash to draw down before touching anything else. In reality it's two separate decisions bundled together. One is how much cash you need this year, which the order answers fine. The other is how much of the account's embedded gain to realize while you're at it, since selling a position at a 0% federal tax rate resets its cost basis for free. That second decision has nothing to do with spending. You can realize far more gain than you need in cash and simply buy the position right back. Whether that's a good idea depends entirely on your tax bracket that year, not on which account category you're withdrawing from.

The same gap shows up on the traditional-account side. Converting traditional dollars to Roth is also a discretionary decision, not a withdrawal. It competes directly for the same bracket room a gain harvest wants. Long-term capital gains stack on top of ordinary income for tax purposes. A bigger Roth conversion in a given year raises the floor gains sit on top of, leaving less of the 0% capital gains bracket available for harvesting that same year. A rule that only tells you which account to spend from can't coordinate two levers pulling on the same limited space. Sizing both, every single year, against that year's actual bracket room is the harder problem. It's the one that moves the tax bill.

Case Study: Ray and Nadia Have Room for Both, but Only in Some Years

Ray and Nadia are 62 and 61, just retired in Texas, with $1.45M combined in traditional 401(k)s and $800,000 in a taxable brokerage account carrying roughly $500,000 of unrealized gain. They delay Social Security to 67, and spend $150,000 a year in the meantime. That six-year gap, before Social Security starts and decades before RMDs, is exactly when their taxable income is at its lowest for the rest of the plan.

Ray and Nadia set their Roth conversion to fill only the 10% bracket, deliberately conservative, not the more common default of filling all the way to 22%. That choice matters here. The top of the 10% bracket sits well below the 0% capital gains ceiling, so after the conversion uses its share, there's still real room left over for harvesting. In 2026 alone, that's a $53,856 harvest, growing to $69,130 by 2029, all realized at a 0% federal rate. A fixed "spend taxable first" rule would draw from this same account for spending. Nothing in it recommends realizing tens of thousands more in gain than the year's cash need, and nothing in it coordinates that against a conversion happening in parallel.

Just as telling is what happens in 2030. The optimal harvest drops straight to $0, the same year Social Security starts. Nothing about their spending changed. What changed is that Social Security now fills the ordinary-income floor that used to be empty, leaving no 0% room for a gain to stack on top of. The right amount to harvest is whatever's left over after everything else that year, not a fixed habit. That number moves every single year of the plan.

Case Study: Owen and Priya Have the Same Setup and Zero Benefit

Owen and Priya are a similar age, also retired in Texas, also holding a taxable account with real unrealized gain, $500,000 built up on a $500,000 balance. The difference is the rest of their portfolio. Most of it already sits in Roth accounts, they're not running a Roth conversion, and they spend a more modest $70,000 a year. A rule that says "harvest gains whenever you're in a low bracket" would look at their income, see it's low, and tell them to start harvesting immediately.

That instinct doesn't hold up here, and Texas, with no state income tax, is the cleanest place to see why. There's no state bill for an early harvest to bring forward in their case, so harvesting is entirely free to attempt. It still accomplishes nothing. Realizing a gain changes only the timing of the tax, not the amount. Owen and Priya would sell this same position at 0% federal tax anyway, whenever they eventually got around to it. Harvesting it today only resets a cost basis they were never going to be taxed on. Zero cost, but also zero benefit, a wash transaction with nothing on either side of the ledger. In a state that taxes capital gains, the same setup would cost something. It would move that state's bill earlier for no reason. Before recommending anything, the optimizer runs their own future years forward first, checking a real question. Is there ever a later year where this same gain would face a higher federal rate than it does today? For this household, the answer is no, in every single year through the end of the plan, so there's nothing to capture by moving early.

The Bottom Line card shows the same honest number from the other direction. $0 converted, $0 harvested, $0 in combined savings versus doing nothing at all. Nothing here failed to find an opportunity. There wasn't one to find, which matters as much as finding the real ones in Ray and Nadia's case above.

The order is a starting point, not the answer

Taxable, then traditional, then Roth remains a reasonable default for which account to draw spending from, and nothing here argues against it. What it was never built to answer is how much gain to realize or how much to convert in a given year. The optimizer resizes those two numbers every year against that year's actual bracket room, and checks them against your own future income before recommending either one. That's where the real difference between households like Ray and Nadia and households like Owen and Priya comes from. Same account types, same rough ages, same state, opposite answers, because the order was never the real question.