Retirement withdrawal-order strategies: which account comes first?

By the RetireGlide Team · August 15, 2026

No single retirement withdrawal order is best for every household. Taxable-first, tax-deferred-first, and proportional withdrawals recognize income and realize gains on different schedules, so a useful answer comes from comparing the same plan under each strategy after required minimum distributions are taken.

RetireGlide engine 1.6.0 models those three choices as educational scenarios. Required minimum distributions (RMDs) are forced first, and Roth remains the last discretionary source in every version 1 strategy.

What does each withdrawal-order strategy mean?

Withdrawal order controls how a remaining cash shortfall is funded after income such as wages, pensions, Social Security, portfolio distributions, and forced RMDs is counted. The strategy changes which account bears the discretionary draw; it does not switch off tax rules or RMD obligations.

The three engine 1.6.0 choices are deliberately narrow so the comparison is understandable and reproducible. They do not include a Roth-first option.

  • Taxable-first: use taxable assets for the discretionary shortfall, then tax-deferred assets, then Roth if the first two sources are exhausted.
  • Tax-deferred-first: use traditional IRA and other pre-tax assets first, then taxable assets, then Roth.
  • Proportional: split the discretionary draw across taxable and tax-deferred balances in proportion to their current balances, with Roth used only after those sources are exhausted.

Why are RMDs taken before the selected order?

RMDs are legal minimum distributions, not optional spending choices. The IRS generally requires owners of traditional IRAs and many employer plans to begin annual distributions at the applicable age, and an RMD is generally included in taxable income except for any previously taxed basis.

Engine 1.6.0 therefore removes the required amount from tax-deferred accounts before applying the selected discretionary order. If the RMD and other income already cover spending and taxes, the selected strategy may have little or no discretionary withdrawal to allocate that year.

How can the three strategies produce different modeled paths?

Consider the same retired household with money in taxable, traditional, and Roth accounts. In a qualitative deterministic comparison, a taxable-first run may realize gains while leaving more pre-tax money for later; a tax-deferred-first run may recognize more ordinary income earlier while reducing the balance exposed to future RMDs; a proportional run spreads those effects across both account types.

Nothing else in that comparison changes: spending, claiming ages, portfolio assumptions, and market-return sequence stay the same. The engine recomputes taxes and the withdrawals needed to pay those taxes together, then reports each strategy's tax path, RMD path, Medicare costs, portfolio balances, and forecast outcomes. This example intentionally gives no numeric winner because no one strategy wins for every household and no deterministic fixture result is being quoted here.

Why does Roth stay last in the version 1 strategies?

Qualified Roth withdrawals generally do not add taxable income, and Roth IRAs do not require lifetime RMDs for the original owner. Keeping Roth last isolates the choice between taxable and tax-deferred funding while preserving Roth as the final discretionary pool in all three comparisons.

That is an engine design boundary, not a claim that Roth should always be spent last. A future Roth-first or threshold-aware strategy would need to be explicitly designed, versioned, tested, and compared as a separate lever rather than implied by an AI explanation.

How should a modeled comparison be read?

Compare more than one output. Lifetime tax can move differently from Medicare premiums, ending assets, annual cash flow, or the size of future RMDs. A strategy that looks favorable on one measure can look less favorable on another, especially when Social Security taxation, capital gains, state tax, or IRMAA enters the path.

RetireGlide's versioned deterministic engine performs every withdrawal, tax, projection, and simulation calculation. AI may propose a strategy input and explain the computed differences, but it never calculates or ranks the outcomes itself; the user confirms any plan input before it is saved.

Frequently asked questions

What is the default withdrawal order in RetireGlide?
Taxable-first is the default when no strategy is selected. RMDs are still taken first, and Roth remains last for discretionary funding in all three engine 1.6.0 strategies.
Does proportional mean equal withdrawals from each account?
No. Engine 1.6.0 divides the discretionary draw between taxable and tax-deferred pools according to their current balances, so the split changes as those balances change.
Can a withdrawal order eliminate RMDs?
No. The selected order applies only after required distributions. Earlier tax-deferred draws can reduce the balance used for later RMD calculations, but the order cannot waive an RMD that is due.
Why can withdrawal order affect Medicare premiums?
Tax-deferred withdrawals and realized taxable gains can change modified adjusted gross income. Medicare generally uses income from two years earlier for IRMAA, so different income timing can produce different premium paths.

See this in your own plan

Run your numbers through the same engine — free, in about 3 minutes.

Sources

RetireGlide is an educational modeling tool, not an investment, tax, or legal adviser. Numbers that change annually (tax thresholds, premiums, benefit formulas) are approximate — always verify against the official sources above. Read our full disclaimer.