The short answer
Historically, mostly yes. A retiree who spent 4% of their savings in the first year, then raised that amount with inflation every year, still had money after 30 years in 122 of the 126 retirements we could test. That covers everyone who retired from 1871 through 1996. It is a record of the past under the assumptions below, not a promise about the next 30 years.
The four that ran out began in 1929, 1966, 1968 and 1969, and even those lasted 25 to 29 years. At 3.5%, none of the 126 ran out in our test.
For most people, 4% turned out to be cautious. The median retirement could have withdrawn up to 7.38% a year and still lasted 30 years, and 88 of the 126 could have managed 6% or more. That is hindsight, not a safe rate. The rule is sized for the bad cases, because nobody knows in advance which kind of retirement they are getting.
Red bars are the years where 4% would have run out. Hover or tap a bar for its rate.
How we ran the numbers
Each retirement starts with a portfolio that is entirely U.S. stocks: the S&P 500, or Robert Shiller’s reconstruction of the broad market before 1957, with dividends reinvested. At the start of every year the retiree takes out 4% of the original balance, adjusted for that year’s inflation, and the rest earns that year’s actual return. If a withdrawal can’t be covered, the retirement has failed.
Stock returns come from Shiller’s data and S&P Dow Jones Indices. Inflation is the Bureau of Labor Statistics’ consumer price index, with Shiller’s estimates before 1913. The data is annual and runs through 2025, so a retirement that began in 1996 is the last 30-year one we can follow to the end. There are no fees, taxes or bonds in these runs; we come back to what that leaves out.
The method follows William Bengen’s 1994 study, which first proposed 4%, and the 1998 “Trinity study”. The portfolio is different: Bengen mixed stocks with intermediate-term Treasury bonds, and the Trinity study tested several mixes from all stocks to all bonds. Ours is 100% U.S. stocks, so the results are comparable but not identical. We also have three more decades of data, and we publish every result.
Success rates at other spending rates
The drop-off is steep. Half a point above 4% turns four failures into eight, and longer retirements punish higher rates hardest. Below 4%, our test found no failures at all: 3.5% lasted through every period of every length from 20 to 50 years.
| Withdrawal rate | 20 years | 30 years | 40 years | 50 years |
|---|---|---|---|---|
| 3% | 100% | 100% | 100% | 100% |
| 3.5% | 100% | 100% | 100% | 100% |
| 4% | 100% | 96.8% | 94.0% | 93.4% |
| 4.5% | 97.8% | 93.7% | 83.6% | 77.4% |
| 5% | 94.9% | 81.0% | 73.3% | 69.8% |
| 6% | 81.6% | 69.8% | 56.9% | 50.9% |
Share of retirements that still had money at the end. Periods tested: 136 of 20 years, 126 of 30, 116 of 40 and 106 of 50.
Try your own numbers
What sank the four that failed
It wasn’t simply a bad market. Over the full 30 years, stocks returned 7.1% a year after inflation for the 1969 retiree, close to their long-run average, and the money still ran out in year 26.
The damage came early. In the first ten years of the 1966, 1968 and 1969 retirements, inflation ran at 5.7% to 6.7% a year while stocks gained only 3.1% to 3.5%. Every withdrawal grew with prices while the portfolio shrank in real terms, and after a decade the 1969 retiree had 36% of their starting balance left. When the long bull market finally arrived in the 1980s, there wasn’t enough left for it to rescue.
1929 failed the other way, with a crash and falling prices. Three years in, the portfolio was down to 39% of its starting value in real terms, and every withdrawal after that came out of a much smaller pot. That gave 1929 the lowest safe rate of any year, 3.78%.
The lesson from all four is the same: what happens in the first decade matters far more than the average over 30 years. Financial planners call this sequence-of-returns risk.
The retiree of 2000 is about to become the fifth
Retirements that started after 1996 haven’t run their 30 years yet, but one is already in trouble. Someone who retired at the start of 2000 with $1 million and spent $40,000 a year, adjusted for inflation, has about $61,500 left in 2000 dollars after 26 years. That is about a year and a half of spending.
Their first decade looks like 1966’s. Two crashes, in 2000–2002 and 2008, left stocks down about 1% a year over ten years before inflation, and after a decade the portfolio was down to 34% of where it began, lower than any of the four historical failures at the same point. Inflation was much milder this time, which is why the money has lasted as long as it has.
A slightly lower rate would have changed the story. At 3.5%, the same retiree would still have about half their starting balance in real terms. At 3%, they would have all of it.
If you’re retiring early
Over 40 and 50 years, 4% held in 94.0% and 93.4% of retirements. The worst 45-year retirements, starting in 1966 and 1969, could sustain only 3.56%. That is why people retiring in their 30s or 40s often plan on 3.25% to 3.5%, or on cutting spending after a bad year.
The other side is easy to miss. At 4%, 105 of the 126 thirty-year retirees finished with more money after inflation than they started with, and the median one ended with three times as much. The rule is sized for the worst stretches in the data, and most of the time it leaves a lot unspent.
What these numbers leave out
Fees. A 1% yearly fee cut the 30-year success rate at 4% from 122 to 118 of 126, exactly the same damage as spending 4.5% with no fee.
Bonds. Our portfolios are all stocks. Bengen’s stock and bond mixes still landed near 4%. Bonds soften early crashes but lower long-run returns.
Taxes. Withdrawals from a traditional IRA or 401(k) are taxed as income, so the 4% has to cover the tax bill too.
American luck. The U.S. had one of the strongest stock markets of the past century. Research on other countries has generally found lower safe rates.
Real behavior. People cut back after a bad year, and Social Security or a pension covers part of their spending. Both make a given rate safer than this test suggests.
Test your own plan
The FIRE calculator runs this same historical test for your withdrawal rate and retirement length, alongside how many years it will take you to get there. To see how a lump sum would have grown over any stretch of years, use the S&P 500 return calculator.
Sources
- Robert J. Shiller, U.S. stock market data (S&P Composite, dividends and CPI), Yale University; S&P Dow Jones Indices for recent years.
- U.S. Bureau of Labor Statistics, Consumer Price Index for All Urban Consumers (CPI-U), December values.
- William P. Bengen, “Determining Withdrawal Rates Using Historical Data,” Journal of Financial Planning, October 1994.
- Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz, “Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable,” AAII Journal, February 1998.
Every figure on this page is computed from the data above by the same code that powers the site’s calculators, and automated tests check that the text matches it. The method is described in full on the methodology page.