Does the 4% rule survive a 40- or 50-year retirement?

By DecisionSheet · Updated September 22, 2026 · 2026 tax figures

The 4% rule takes 4% of a portfolio in the first year of retirement, then withdraws that same dollar amount every year after, raised with inflation. The Trinity study, which tested it against history, covered retirements of up to 30 years. Early retirement asks it to last 40 or 50.

The case here is a $1,500,000 portfolio, 25 times spending, withdrawing $60,000 a year in today's dollars from 80% stocks and 20% bonds. The only thing that changes is the horizon. The success rate falls from 95.9% at 30 years to 91.3% at 40 and 88.6% at 50, while the median ending balance rises from $3,783,484 to $9,897,064.

What "survive" means here

The Trinity study (Cooley, Hubbard and Walz, AAII Journal, February 1998) ran withdrawal rates from 3% to 12% through every overlapping payout period of actual US stock and corporate bond returns from 1926 to 1995. It counted a period as a success if the portfolio ended above $0. The longest payout period it tested was 30 years. With withdrawals adjusted for inflation, a 4% rate succeeded in 98% of those 30-year periods for a 75/25 stock/bond mix and 95% for all stocks. The study did not test 40 or 50 years.

DecisionSheet's FIRE simulator asks the same question of simulated markets rather than historical ones. Each of 1,000 runs withdraws the year's spending at the start of the year, then applies one random annual return. A run fails the year its balance reaches zero. The success rate is the share of runs still above zero at the end of the horizon. Spending and balances are in real, inflation-adjusted dollars throughout, so "$60,000 a year" means the same purchasing power in year 50 as in year 1.

The three horizons

Open each row: 30 years · 40 years · 50 years. The three scenarios differ only in the horizon.

Across the three horizons, both sides of the distribution move outward. The median ending balance at 50 years is $9,897,064 in today's dollars, 6.6 times the starting balance. Meanwhile more than one run in ten is empty by year 50.

The simulator seeds its random numbers from the market assumptions alone, and each run has its own stream of annual returns. So all three links run on the same 1,000 simulated markets, and the 30- and 40-year results are the first 30 and 40 years of the 50-year paths. The rows differ because the same runs keep failing later, not because each row drew different markets: in this draw, 46 of the 1,000 runs ran out in the fourth decade and 27 in the fifth. The first failure came in year 16.

The level of each figure does depend on the draw. DecisionSheet simulated the same assumptions under 10 other seeds (runs of 1,001 to 1,010, since the run count is part of the seed). Across those and the linked draw, the 30-year success rate ranged from 93.8% to 95.9%, the 40-year from 88.3% to 91.7% and the 50-year from 84.7% to 88.6%. The linked draw is among the more favorable: its 30- and 50-year rates are the highest of the 11 draws, and its 40-year rate the second highest. The rows kept their order on every draw, as shared paths guarantee.

The 10th percentile

The fan chart plots three percentiles of the 1,000 balances each year. The 10th percentile is the balance that 90% of runs are at or above that year. The chart's vertical scale is set by the 90th percentile, so at 50 years the 10th-percentile line sits close to the axis.

In the 50-year run the 10th-percentile line is at $607,390 in year 30 and reaches $0 in year 44. Once more than one run in ten has run out, the 10th percentile can only be zero. That is why the table shows $0 at 50 years. At 40 years the 10th percentile is $224,866 on this draw but $0 on 5 of the 11. Once the 10th percentile is zero, the success rate is the more informative figure.

Each point on the line is ranked separately, year by year. The line does not follow any single market: the run at the 10th percentile in one year need not be the one there the next. The calculator labels the line "sequence risk", meaning the risk that poor returns arrive early, while withdrawals are selling a larger share of a smaller portfolio. The simulation contains such sequences because every year's return is drawn independently. But it cannot isolate them, so the chart shows how wide the outcomes are rather than why any one run failed.

Why the inputs are assumptions

Every figure above rests on assumed returns, not forecasts. They are the calculator's defaults: stocks return 6.8% a year after inflation with 16.0% volatility (the standard deviation of annual returns), and bonds 2.2% with 6.0%. The return inputs are compound annual growth rates. The simulator converts each to an approximate average annual return by adding half its variance before drawing, because a series of random annual returns compounds to less than its average. At a 5.8% stock return instead, every success rate falls below its range across draws. The model's answer to "does 4% survive 50 years" is conditional on these numbers, and the calculator lets each one be replaced.

Assumptions and limits

• Annual returns are independent draws from a normal distribution, with a fixed stock/bond correlation of 0.15 and annual rebalancing. Nothing carries over from one year to the next, so long runs of bad years arise only by chance, and extreme market years may be underrepresented.
• Returns and spending are real. Inflation is not modeled separately, so a high-inflation decade is not a distinct risk here.
• No fees, no taxes on withdrawals, no Social Security, pension or other income. There is no input for any of them. Every dollar of spending comes from the portfolio for the whole horizon.
• The withdrawal is fixed in real terms and never adjusts to the portfolio. The calculator's other two withdrawal rules do adjust, and change these figures.
• The horizon is fixed. The model does not estimate how long anyone lives.
• Success counts any run above $0 at the end. A run that ends with a few dollars counts the same as one that ends with $9,897,064.
• 1,000 runs pin a success rate down only so far: across the draws above it moved by up to 3.9 percentage points.

Method and sources

The model is calculateFire in the FIRE & safe withdrawal calculator, run with the calculator's default allocation, returns and volatilities, a fixed real withdrawal strategy, and a 4% safe withdrawal rate. The data file checks that the three horizons share their paths and that annual savings changes nothing here.

• Cooley, Hubbard and Walz, "Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable", AAII Journal, February 1998: the 1926–1995 data, payout periods of 15 to 30 years, success defined as an ending value above $0, and Table 3's inflation-adjusted success rates of 98% (75/25) and 95% (100% stocks) at 4% over 30 years.
• "Multi-period Returns", Macro-Investment Analysis (Stanford): the approximation that the geometric return is the arithmetic mean less half the variance, which the simulator uses in reverse.

Open this scenario in the calculator

All figures on this page come from the FIRE Simulator calculator. Change any input there and the numbers update.