Deciding how to pull money out of a retirement portfolio decides more of your outcome than most savers believe. Two retirees with identical balances can land decades apart, and the difference is almost always the withdrawal plan. This guide covers the 4% rule and where it breaks, guardrail strategies that adapt to markets, the order in which accounts should be spent, Roth conversion ladders, sequence-of-returns risk, and the required minimum distribution rules in force for 2026.
The rule has a precise origin. In 1994, financial planner William Bengen published research in the Journal of Financial Planning asking a blunt question: what is the highest first-year withdrawal, later adjusted for inflation, that would have survived every 30-year retirement in US market history? His answer for a portfolio split roughly evenly between stocks and bonds was about 4%. In 1998, three professors at Trinity University (Philip Cooley, Carl Hubbard, and Daniel Walz) ran the same experiment across more historical periods and reached a similar range, and the "Trinity study" cemented 4% as the number everyone quotes.
The strength of that research is that it ignores hindsight. It did not pick a friendly sample. It forced the portfolio to survive the 1970s inflation, the 1929 crash, and the stagflation of the early 1980s, all while the retiree kept withdrawing the same inflation-adjusted amount. That worst-case framing is why the rule became so durable.
The attacks on it are equally specific, and we think most of them land. First, the horizon assumption: 30 years fits someone retiring at 65, and it fits badly for a 45-year-old who may need 45 or 50 years of funding. Second, the data: Bengen and the Trinity authors tested US returns only, and researchers such as Wade Pfau have shown that international markets produced materially lower safe withdrawal rates over comparable periods. A rule calibrated on the single best-performing market of the twentieth century carries survivorship bias. Third, the behavior assumption: the rule requires you to spend the same inflation-adjusted amount in year 25 as in year 1, through every crash. Almost nobody lives that way, which means the historical success rate both overstates the risk for flexible retirees and understates it for rigid ones.
Industry research has moved accordingly. Fidelity's retirement guidance suggests holding initial withdrawals to roughly 4% to 5% of the starting balance, adjusted for inflation, while stressing that the sustainable number depends on horizon, allocation, and market conditions. Morningstar's annual "State of Retirement Income" research has put its base-case starting rate somewhere between about 3.7% and 4.0% across recent editions. The honest summary: 4% sits near the center of a defensible range for a 30-year horizon with a balanced portfolio, and outside that range the number changes fast.
Fixed-percentage-of-balance withdrawals (spending 4% of whatever the portfolio holds each year) can never run out of money, but they produce volatile income. Inflation-adjusted fixed-dollar withdrawals produce stable income but can run out. Guardrail strategies sit between the two and, in our view, represent the best practical compromise most retirees can actually follow.
The best-known version comes from Jonathan Guyton and William Klinger's 2006 papers. You pick an initial withdrawal rate, often 5% to 5.5% for a 30-year horizon, higher than a strict 4% rule would allow. Then you install two rules. If the current year's withdrawal rate rises 20% above your initial rate, a "capital preservation" rule cancels that year's inflation adjustment. If the rate falls 20% below the initial rate, a "prosperity" rule grants an extra inflation bump. Research from Vanguard on dynamic spending strategies reaches the same conclusion from a different direction: strategies that cut spending after poor market results support meaningfully higher starting income than rigid rules, for nearly identical failure probabilities. Vanguard's own commentary has long emphasized that no single withdrawal rate suits everyone, and that flexibility is what buys the extra income.
The trade is real and you should see it clearly: guardrails front-load spending when you are healthiest and most able to enjoy it, at the price of a smaller paycheck in bad years. Retirees who cannot tolerate any income cut should start at a lower fixed rate instead. Retirees who can accept occasional cuts get a higher standard of living most years. The Retirement Withdrawal Calculator models a fixed real withdrawal schedule year by year, which is the right baseline before you layer guardrail adjustments on top.
Accumulators care about average returns. Withdrawers care about the order those returns arrive in. The arithmetic behind that claim is small enough to check by hand: a portfolio that gains 10% and then loses 10% ends at 99% of its starting value, even though the average return was zero. A retiree withdrawing 4% a year from that same two-year stretch ends noticeably lower, because the second withdrawal comes out of a smaller base.
Now scale that to a retirement. A 10% portfolio gain followed by a 10% loss, with 4% of the initial balance withdrawn each year, leaves roughly 91% of starting capital. Reverse the order and the result lands within a percentage point, so a single mild cycle barely matters. What matters is a deep bear market early. Take the same retiree and give them a 30% loss in year 1: a 4% withdrawal taken at the bottom permanently removes shares that a 43% rebound would have restored. Research from both Fidelity and Morningstar has repeatedly found that poor returns in the first five years of retirement do more damage to portfolio longevity than the same poor returns later, because later losses hit a portfolio that has already been drawn down by years of spending.
Three defenses actually work. A cash reserve of one to two years of planned withdrawals, rebuilt in good years, lets you skip selling during a crash entirely. Flexible spending does the same job implicitly, since cutting a withdrawal by 10% in a down year removes far less permanent capital than it appears to. And starting retirement at a withdrawal rate with slack, rather than the highest rate historical tables allow, converts sequence risk from a threat into an inconvenience. There is no investment product that reliably removes sequence risk; anyone selling one is charging you for the promise.
The conventional order is simple: spend taxable accounts first, tax-deferred accounts (traditional IRA, 401(k)) second, Roth last. The logic is sound. Taxable money is already taxed, so spending it costs nothing in sheltered space. Traditional money grows tax-deferred and will be taxed at withdrawal no matter what, so leaving it invested as long as possible maximizes tax-free compounding. Roth money is the most valuable dollar in the portfolio, untaxed forever and free of required distributions, so it compounds last and longest.
We support that default, and we also think it gets followed more mechanically than it deserves. The order is a tax optimization, and tax brackets change every year. Three situations routinely justify breaking it.
The empty-bracket problem. A retiree living mainly on taxable savings may report taxable income below the 2026 standard deduction of $16,100 single or $32,200 married filing jointly (IRS Rev. Proc. 2025-32). Those brackets expire unused every year they stay empty. Withdrawing from the traditional IRA to fill the 10% and 12% brackets, which top out at $50,400 and $100,800 of taxable income for single and joint filers, converts money that would have been taxed at 22% or 24% in later life at a fraction of the price.
The surcharge cliff problem. Medicare's IRMAA surcharges are set from modified adjusted gross income with a two-year lookback, and up to 85% of Social Security benefits become taxable once other income rises (IRS Pub 915). A retiree who parks a seven-figure traditional balance for 20 years and then faces RMDs stacked on top of Social Security can end up paying more lifetime tax than a retiree who drew the IRA down deliberately through the low-income years, even at some cost to Roth compounding.
The bracket-waste problem in reverse. In a year with unusually low income, such as a sabbatical or the year before Social Security starts, the cheapest tax you will ever pay on retirement money is available right then. Converting traditional dollars to Roth in those years, at rates that can fall below 10% effective, is often worth more than any amount of perfect ordering among accounts.
The practical rule we suggest: treat the conventional order as your starting point, then check the current year's brackets before every withdrawal decision. The money you spend is rarely the interesting question; the taxable income the withdrawal creates is.
Retiring before 59 and a half introduces a problem the conventional guides skip: most withdrawals from retirement accounts before that age carry a 10% early-distribution penalty. The standard escapes are narrow. The Rule of 55 allows penalty-free withdrawals from the 401(k) of the employer you just left, if you separate from service in or after the year you turn 55. Substantially equal periodic payments under IRC Section 72(t) lock you into fixed withdrawals for five years or until 59 and a half. Both are rigid, and neither is a plan.
A Roth conversion ladder is the flexible alternative. Conversions from a traditional IRA to a Roth have no age minimum and no penalty at any age. In each low-income year of early retirement, you convert the amount that fills your target bracket, pay the tax from savings held outside the retirement accounts, and wait five years; after that, each converted amount and its earnings can be withdrawn from the Roth tax-free and penalty-free. Year by year, the ladder converts tomorrow's forced taxable income into today's cheap taxable income.
The strategy has honest failure modes. A stagnant market, a short conversion window, or a modest balance can leave the conversion taxes buying nothing, which is why the ladder deserves to be priced before it is run rather than assumed to win. The Roth Conversion Ladder Calculator runs the full year-by-year comparison against the 2026 brackets, including the years where the ladder loses, and the single-shot version of the same decision is covered in the Roth Conversion Calculator. And before 59 and a half, any withdrawal taken outside the ladder still triggers the penalty, which the 401k Early Withdrawal Calculator prices directly.
Every strategy above runs into the same wall eventually. Under SECURE 2.0, owners of traditional IRAs and most workplace plans must begin taking required minimum distributions at age 73 for anyone born 1951 through 1959, and at 75 for anyone born in 1960 or later, per the IRS retirement plan and IRA RMD rules. Roth IRAs carry no RMDs during the original owner's lifetime, and Roth 401(k) accounts have been exempt since 2024.
The mechanics are unforgiving in specific ways. The first RMD covers the year you reach the trigger age and can be delayed until April 1 of the following year, though delaying means two taxable distributions in one year. Each year's amount equals the prior December 31 balance divided by a life expectancy factor from the IRS Uniform Lifetime Table in Publication 590-B; at 73 the divisor is 26.5, about 3.8% of the balance, and the percentage climbs every year. Miss an RMD and the shortfall faces a 25% excise tax, reduced to 10% if corrected within two years, per the IRS RMD FAQ. These are ordinary-income dollars whether you need them or not, which is exactly why the tax planning in this guide concentrates on the years before the wall arrives.
For guardrail strategies, RMDs change the frame rather than break it. After the trigger age, the RMD can serve as the floor of the withdrawal plan, with guardrail adjustments applied to spending above it. The RMD Calculator projects the year-by-year forced distributions from any balance, and the Roth conversion ladder exists in large part to shrink that floor before it arrives.
Withdrawal strategy and claiming strategy are one decision viewed from two sides. Delaying Social Security past full retirement age raises the benefit by roughly 8% per year of delay up to 70, a guaranteed, inflation-adjusted increase no portfolio can match. But the years of delay must be funded from the portfolio, which means larger withdrawals exactly when the sequence-of-returns exposure is highest. Claiming early does the reverse: it protects the portfolio in the fragile first years at the cost of a permanently smaller benefit.
For most retirees with adequate savings, delaying the higher earner's benefit to 70 and bridging the gap with portfolio withdrawals, sized to stay inside the low brackets, is the better trade. The break-even age where the delayed benefit overtakes the early one typically lands in the low 80s, and the Social Security Break-Even Calculator locates it for any specific pair of benefits. One caution worth repeating: taxable Social Security interacts with IRA withdrawals, so every claiming change reshapes the optimal withdrawal order in the years around it.
Condensing everything above into a plan: start between 4% and 5% of the initial portfolio depending on your horizon, closer to 4% if retirement could last 40 years or more. Install guardrails rather than promising yourself a fixed inflation raise forever, and hold one to two years of withdrawals in cash so bad markets never force a sale. Follow the taxable-then-tax-deferred-then-Roth order as a default, and break it whenever the current year's brackets say a traditional withdrawal, a conversion, or a Roth draw is cheaper. Use the gap years between retirement and Social Security to convert at low rates. And treat the RMD age, 73 or 75 depending on birth year, as the deadline by which the tax geography of your accounts should already be arranged, because after that the IRS starts making the withdrawals for you.
It is a reasonable starting point and a poor finish line. William Bengen's 1994 research and the 1998 Trinity study found that withdrawing 4 percent of the initial balance, adjusted for inflation, survived every 30-year period of US market history they tested, for a portfolio split roughly evenly between stocks and bonds. The assumptions behind that survival rate are the weak spot: a fixed 30-year horizon, US-only data, constant inflation-adjusted spending in every year, and historical valuations that were often lower than today's. Modern research, including Fidelity's guidance to hold initial withdrawals between 4 and 5 percent and Morningstar's annual updates, treats the sustainable starting rate as a moving target. Treat 4 percent as the center of a range, then add flexibility.
The textbook order is taxable first, tax-deferred second, Roth last, because it lets the tax-sheltered accounts compound the longest. It is a good default and a bad law. Spending down a traditional IRA first can be smarter when your taxable income is so low that the standard deduction goes unused, or when keeping the IRA balance high would push you over a Medicare IRMAA surcharge or increase the taxable share of Social Security a few years later. The right framing is that the order is an annual tax-bracket optimization, and the goal is to fill the cheapest brackets you will ever have access to, which sometimes means converting instead of withdrawing.
You start with a higher initial withdrawal rate than the 4 percent rule allows, then adjust only when the portfolio crosses a limit. In the Guyton and Klinger framework, if the current withdrawal rate rises 20 percent above the initial rate, you skip the inflation raise that year, and if it falls 20 percent below, you take a raise. Vanguard's research on dynamic spending finds that rules that cut spending after bad markets support meaningfully higher starting income for nearly the same failure risk. The cost is behavioral, not mathematical: you have to actually accept a smaller paycheck in a bear market.
In the gap years between retirement and required minimum distributions, when your taxable income is lowest. Retiring before 59 and a half adds a penalty problem, which the ladder solves by converting each year rather than withdrawing; conversions from a traditional IRA to a Roth have no age limit and no penalty. Each conversion should be sized to the top of a low tax bracket, with the tax paid from savings outside the retirement accounts. Our Roth conversion ladder calculator runs the year-by-year math against the 2026 brackets.
Under SECURE 2.0, you must begin required minimum distributions from traditional IRAs and workplace plans at 73 if you were born 1951 through 1959, and at 75 if you were born in 1960 or later. The first RMD can be delayed to April 1 of the year after you reach the age, though delaying means two distributions in one tax year. Each year's amount is the prior December 31 balance divided by a life expectancy factor from the IRS Uniform Lifetime Table in Publication 590-B. Miss one and the excise tax is 25 percent of the shortfall, reduced to 10 percent if corrected within two years. Roth IRAs have no RMDs while the original owner is alive.
While you are withdrawing, yes. Two portfolios with identical average returns can end at wildly different balances depending on when the losses arrive, because a withdrawal taken during a downturn permanently removes shares that would have participated in the recovery. A 10 percent loss followed by a 10 percent gain does not return you to even; it leaves you 1 percent down before any spending. The practical defenses are a cash reserve covering one to two years of withdrawals, a willingness to trim spending after bad years, and starting retirement with a withdrawal rate below what the averages suggest you could afford.
Every strategy in this guide can be priced before you commit to it. The Retirement Withdrawal Calculator tests a full year-by-year withdrawal schedule against your own balance, returns, and inflation. The Roth Conversion Ladder Calculator values the gap-year conversion window. The RMD Calculator projects the forced distributions that arrive at 73 or 75. The Social Security Break-Even Calculator locates the claiming age where delaying pays off. And the FIRE Calculator works back from a target retirement date to the balance the whole plan depends on.
Sources: IRS, Retirement Plan and IRA Required Minimum Distributions FAQs and Publication 590-B (RMD ages, April 1 rule, Uniform Lifetime Table, 25%/10% excise tax); IRS Rev. Proc. 2025-32 (2026 tax brackets and standard deductions); IRS Pub 915 (Social Security taxation); William Bengen, Journal of Financial Planning, 1994; Cooley, Hubbard and Walz, Trinity University, 1998; Guyton and Klinger, Journal of Financial Planning, 2006; Fidelity Viewpoints retirement research (4% to 5% initial withdrawal guidance); Vanguard dynamic spending research and Morningstar State of Retirement Income (sustainable-rate estimates).
All calculation logic on this page is implemented from the primary documents below. Each one is listed with the figure it supplies, so any rate or threshold in the result can be traced back to its published source.
Benefit rules and cost assumptions stated on the page, not implied.