4% vs 3.5% SWR for Early Retirees: Horizon Changes the Math

The original work looked at a roughly 30-year retirement, using historical U.S. stock and bond sequences. Early retirees often plan for 40 to 50 years of withdrawals. That extra horizon is the honest reason many people start at 3.5% or 3% instead of 4% — not because a new study overturned the old one, and not because a calculator proved a winner.

A withdrawal rate is a starting assumption to stress-test. It is not a stopping point.

What the 4% Rule Actually Said

In October 1994, William P. Bengen published "Determining Withdrawal Rates Using Historical Data" in the Journal of Financial Planning. He reconstructed retiree experience with historical U.S. returns and inflation. He did not use Monte Carlo. He used rolling historical periods.

The finding that became the "4% rule" was specific. For a retirement that needed to last at least 30 years, Bengen wrote that "a first-year withdrawal of 4 percent, followed by inflation-adjusted withdrawals in subsequent years, should be safe." He counseled clients to withdraw at no more than 4% in that 30-year framing.

Two other sentences from the same paper usually get dropped:

• At 4%, "In no past case has it caused a portfolio to be exhausted before 33 years, and in most cases it will lead to portfolio lives of 50 years or longer."
• A 4.25% first-year withdrawal "could exhaust a portfolio in as little as 28 years, were past conditions to repeat themselves."

So 4% was the 30-year line he was willing to call safe — not a claim that every rate near 4% survived every past sequence, and not a claim that 4% was right for every horizon. He never named it "the 4% rule." That shorthand came later, and it strips out the horizon, the inflation adjustment, and the fact that this was a worst-case reading of past U.S. sequences, not a forecast.

What the Trinity Study Added

The other paper people mean when they say "the 4% rule" is the Trinity study: Philip L. Cooley, Carl M. Hubbard, and Daniel T. Walz, "Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable," AAII Journal, February 1998.

The authors used rolling historical U.S. stock and bond data (roughly 1926–1995). They looked at several stock/bond mixes and a range of withdrawal rates. A portfolio counted as a success if it still had money at the end of the payout period.

Two limits are easy to miss:

1. The payout periods were 15, 20, 25, and 30 years. Not 40. Not 50.
2. Success was defined against those historical U.S. sequences. Future markets are not required to replay 1926–1995.

For inflation-adjusted withdrawals, they wrote that rates of 3% to 4% "continue to produce high portfolio success rates for stock-dominated portfolios." That is a qualitative finding from a 30-year-or-less historical study. It is not a success-rate table for a 45-year FIRE plan, and this article will not invent one.

Later updates of Trinity do not always agree with each other or with casual recaps online. If you need a historical success percentage, read the original table and the exact conditions — mix, inflation adjustment, and horizon. Our calculator does not validate or refute Trinity.

Why a 30-Year Study Is a Weak Fit for a 45-Year FIRE Plan

Traditional planning often assumes someone stops work around 65 and needs about 30 years of withdrawals. FIRE is a different problem. Retire at 40 and you may be withdrawing for 45 or 50 years.

Two things change when the clock runs that far. Sequence of returns risk — poor markets early, while you are selling shares to live — has more calendar time to compound. A plan that survives 30 years can still fail in year 38. And the 4% rule is not "take 4% of the current balance every year." It is take 4% in year one, then raise that dollar amount with inflation. A long FIRE retirement is a long series of those raises.

Bengen addressed the longer horizon directly. For roughly 50-year longevity, he wrote that an "absolutely safe" initial withdrawal — "to the extent history is a guide" — is 3 percent, "in that it ensures that portfolio longevity is never less than 50 years." He added that this was "also true for withdrawal rates as high as approximately 3.5 percent."

That is the spine of the 4% vs 3.5% debate for early retirees, and it is in the 1994 paper. History is still only a guide. But if you are using Bengen as the reason you picked 4%, it is only fair to use Bengen as the reason a 45- or 50-year plan often starts lower.

4% vs 3.5%: What Changes in the Math

A safe withdrawal rate and a FIRE multiple are the same identity written two ways: FIRE number = annual expenses ÷ SWR, or multiple = 1 ÷ SWR. That is arithmetic, not a success rate.

Factor	4% SWR	3.5% SWR	3% SWR
FIRE multiple	25x expenses	About 28.6x	About 33.3x
On $40,000 spending	$1,000,000	About $1,143,000	About $1,333,000
On $60,000 spending	$1,500,000	About $1,714,000	$2,000,000
Extra capital vs 4%	—	About 14% more	About one-third more
Bengen 1994 framing	"Should be safe" for a 30-year minimum life	About 3–3.5% for 50-year longevity, history as a guide	"Absolutely safe" for 50 years, to the extent history is a guide
Trinity 1998 horizon	Payouts studied up to 30 years	Same 30-year cap	Same 30-year cap

Moving from 4% to 3.5% is not a small tweak. You need about 14% more portfolio for the same spending — or you spend $35,000 instead of $40,000 from a $1,000,000 stash. None of those multiples say the plan "worked X% of the time." They only say how large a stash the rate implies.

If you are still accumulating, the extra 14% is a savings and time problem. If you are close to the 25x line and itching to quit, it is a lifestyle problem: work longer, spend less, earn a bridge income, or accept that a bad early sequence may force a change.

Sequence of Returns Still Sits Underneath Any SWR

Any constant withdrawal rate assumes you keep selling through good years and bad. The average return over 40 years can look fine on a spreadsheet and still fail if the bad years arrive first.

That is sequence of returns risk. A 25% drop in year 1 of retirement is not the same event as a 25% drop in year 20. Early on, you are selling more shares at low prices, leaving fewer shares to recover.

Bengen and Trinity were attempts to respect that problem by looking at actual historical orderings instead of a flat 7% every year. A 4% or 3.5% starting rate is a blunt instrument aimed at the same issue: leave enough room that a bad opening decade does not end the plan.

Blunt instruments have limits. They do not know your cash buffer, your ability to cut spending, or a crash in year 1 versus year 12. Use the Sequence of Returns Calculator to apply a crash of a chosen size in year 1, 5, or 10 and compare it with your baseline. A lower SWR reduces how much you must sell in a down year. It does not eliminate sequence risk.

Why a Single Rule Is the Wrong Stopping Point

Pick 4% or 3.5% and you have chosen a starting withdrawal, not a proof. Hold expenses constant, change the rate, and see what breaks.

Bengen used historical rolling periods. Trinity did too, on a 15- to 30-year grid. Those answer: "Would this withdrawal have survived past U.S. sequences of this length?" If that is your question, tools like FIRECalc and ProjectionLab replay historical sequences. We do not. We do not claim to.

The question we can help with is different: given the expected return and volatility you type in, how often do randomized paths still have money left? That is parametric Monte Carlo — Gaussian draws (Box–Muller) around your assumptions, 100 to 1,000 runs, with percentile bands. Probability of success here means the share of simulations that end with money remaining. It is not a Trinity-style historical success table, and it is not a claim that 4% or 3.5% "works."

Use a rule to size the first draft of a FIRE number. Use a simulation to see whether that draft is fragile.

What Our FIRE Calculator Does

My Projection Calculator is a free, privacy-first FIRE / SWR tool. No signup. It is not a full replacement for a paid planner, and it does not model taxes, Roth conversions, or historical backtesting.

You enter a target portfolio and expected annual expenses, choose a withdrawal rate, and see the implied FIRE number. The on-page FAQ already notes that for 40–50 year horizons, many planners suggest a more conservative 3–3.5% starting rate. You can test rates from conservative (around 3%) to aggressive (around 5%). At 4%, the number is 25× expenses. At 3.5%, about 28.6×. At 3%, about 33×.

Then take the same expenses into the Monte Carlo retirement calculator and see a range of outcomes — 10th through 90th percentile bands, plus the share of paths with money left — instead of one smooth line. The main app puts the same tools on one page.

What it does not do: replay Shiller or Ibbotson sequences, output a Trinity success percentage, or tell you that 3.5% is "correct" for FIRE. Do not treat a single SWR from our screen as a research finding.

Common Mistakes to Avoid

Mistake 1: Treating 4% as a law

Bengen's 1994 result was a 30-year, inflation-adjusted, historical-U.S. finding, rounded into a slogan. Using the slogan without the horizon is how people retire at 38 on 25x and call it "the 4% rule." The paper's own 50-year discussion sat closer to 3–3.5%.

Mistake 2: Using a 30-year study for a 45-year plan and calling it conservative

"The Trinity study said 4% is fine" skips the payout-period column. Trinity's published grid stopped at 30 years. A 45-year FIRE plan is outside that grid. Longer horizon, more sequence risk, more inflation-adjusted withdrawals — that is a reason to start lower or to test harder, not a reason to feel extra safe.

Mistake 3: Confusing 25x with "this lasts forever"

25x is 1 ÷ 0.04. 28.6x is 1 ÷ 0.035. Those multiples do not contain a success probability. A portfolio can be 25x expenses and still fail if early returns are poor and withdrawals stay fixed.

Mistake 4: Stopping at the rule instead of stress-testing

Choosing 3.5% over 4% buys margin in the starting rate. It does not show you a year-1 crash or a 10th-percentile Monte Carlo path. The rate is the assumption. The test is the work.

Mistake 5: Mixing up Monte Carlo with historical success rates

Parametric Monte Carlo (what we run) draws random returns around the expected return you enter. Historical replay (FIRECalc, ProjectionLab's backtesting) asks whether past market sequences would have funded the plan. A high success rate in Gaussian simulations is not Trinity. If someone quotes a success percentage without naming the method, mix, horizon, and inflation rule, treat the number as unfinished.

How to Test 4% and 3.5% With Your Numbers

Keep expenses constant so the only thing that moves is the rate.

1. Write down annual spending in retirement — including health insurance if you will not have an employer plan. Understated expenses make every SWR look safer than it is.
2. Open the FIRE SWR calculator. Enter those expenses and a 4% withdrawal rate. Note the FIRE number (25x).
3. Change only the rate to 3.5%. Note the new number (about 28.6x). That gap is the extra capital — or extra working years — a more conservative start requires. Optionally try 3% (~33x).
4. Take the same expenses into the Monte Carlo calculator. Set the horizon to the years you actually need (40–50 for many early retirees, not 30). Run 100–1,000 paths. Look at the percentile bands and the share of simulations with money left. Then change the withdrawal rate and run it again. Do not stop at a rule of thumb.
5. If a bad first decade is what you worry about, run the sequence of returns stress test with a crash in year 1, then again in year 5 or 10.

Example: $50,000 of annual expenses is $1,250,000 at 4%, about $1,429,000 at 3.5%, and about $1,667,000 at 3%. That is the sizing step. The planning step is whether that stash, withdrawing an inflation-adjusted amount, still has money across noisy return paths for your number of years.

Frequently Asked Questions

Is the 4% rule still valid for early retirees?

It is valid as a 30-year, historical-U.S. rule of thumb, which is what Bengen offered in 1994: a 4% first-year withdrawal, then inflation-adjusted withdrawals, "should be safe" for a 30-year minimum life. Early retirees often need 40–50 years. In that same paper, Bengen described 3% as "absolutely safe" for 50-year longevity to the extent history is a guide, and said that was also true up to about 3.5%. Many FIRE planners therefore start at 3–3.5% and treat 4% as aggressive for a very long horizon.

Why use 3.5% instead of 4%?

Because the extra 10–20 years of withdrawals add sequence risk and more inflation-adjusted spending. Bengen's 50-year historical reading sat around 3–3.5%, not 4%. The cost is a larger stash: about 28.6× expenses instead of 25×, or lower spending from the same portfolio. 3.5% is still a starting rate to test, not a new law.

How much more do I need at 3.5% than at 4%?

About 14% more. $40,000 of spending implies $1,000,000 at 4% and about $1,143,000 at 3.5%. $60,000 implies $1,500,000 vs about $1,714,000. At 3%, the multiple is about 33×. These are 1 ÷ SWR identities, not historical success rates.

Does your calculator prove that 4% or 3.5% will work?

No. The FIRE calculator converts expenses and a chosen SWR into a target number. The Monte Carlo tool then runs 100–1,000 Gaussian simulations around the expected returns you enter and reports the share of paths with money left, plus percentile bands. That is not a historical success table, and it is not a validation of Bengen or Trinity. We do not do tax modeling, Roth conversions, or historical backtesting.

Where can I see historical sequence success rates?

Not here. Bengen and Trinity used rolling historical U.S. sequences; FIRECalc and ProjectionLab are the usual tools if you want that style of replay ("what if I had retired in 1966?"). Our Monte Carlo is parametric — random draws around your assumptions — which is a different method. Use historical replay for the historical question, and Monte Carlo for the "noisy around my inputs" question.

The Bottom Line

4% and 3.5% are not rival religions. They are two starting withdrawals with two implied stash sizes, drawn from research that was always horizon-specific:

• 4% (25x expenses) is the 30-year line Bengen was willing to call safe in 1994, and the figure FIRE turned into a slogan. In his historical cases, 4% did not exhaust a portfolio before 33 years, and often lasted 50 years or longer — in the past U.S. data he studied.
• 3.5% (~28.6x) sits with Bengen's longer-horizon discussion: for 50-year longevity, he treated about 3–3.5% as the "absolutely safe" band, to the extent history is a guide. That extra ~14% of capital is the price of starting more slowly on a 40–50 year plan.
• Trinity (1998) added historical success rates for stock/bond mixes, but only out to 30-year payouts. It does not fill in a 45-year table for you.

A 30-year study is a weak fit for a 45-year FIRE plan. That is the argument. It does not require a fake success-rate chart.

Do not stop at the slogan. Plug the same expenses into 4% and 3.5% on the FIRE calculator, then run both through Monte Carlo. If the question you actually care about is historical sequences, use a tool that replays them. If the question is whether your numbers look fragile once returns wander, that is the test we built.