The 4% rule says you can take 4% of your portfolio in the first year of retirement, increase that dollar amount by inflation every year afterwards, and not run out of money for thirty years.
The second half is where most people get it wrong. It is not 4% of the balance each year. It is 4% once, at the start, and then a fixed inflation-adjusted income for life. On a $1,000,000 portfolio the first year's income is $40,000. At 3% inflation the second year's is $41,200, and it is $41,200 whether the portfolio has grown to $1,100,000 or fallen to $800,000.
Those are two entirely different withdrawal strategies with two entirely different risks. Taking 4% of the balance annually can never exhaust the portfolio, because 4% of a shrinking number is a shrinking number, but it can cut your income in half in a bad decade. The rule as written does the opposite: it holds your income steady and puts all of the risk on the portfolio.
Where the number came from
William Bengen, a financial planner, published it in the Journal of Financial Planning in October 1994. He took US market data from 1926 to 1992 and ran every thirty-year period in it: a retiree starting in 1926, one starting in 1927, and so on. For each starting year he asked what the highest first-year withdrawal rate could have been without the portfolio running dry inside thirty years.
Most starting years tolerated far more than 4%. A 1975 retiree could have taken well over 6% and died rich. But Bengen was not looking for the average. He was looking for the worst one, and the worst ones were the retirees who started in the mid to late 1960s and walked straight into the 1970s: high inflation, a long flat market, and withdrawals rising every year to keep pace.
The worst case in his data survived on 4.15%. He called that the SAFEMAX, rounded it down to 4%, and that is the whole derivation. The Trinity study, published by three Trinity University professors in 1998, ran a similar exercise across different portfolio mixes and time horizons and reached compatible conclusions, which is why the two are usually cited together.
What "safe" meant in the study
It meant the balance was above zero at the end of year thirty. Nothing else.
It did not mean the retiree was comfortable. In several of the surviving runs the portfolio spent years down 60% or more from its starting value while the withdrawals kept rising with inflation, and the retiree had no way of knowing at the time whether they were in a run that recovered or one that did not.
It did not mean anything was left over. A run that finished with one dollar counted as a success on the same terms as one that finished with four million.
And it did not mean thirty-one years. The horizon was a fixed input, not a finding.
Five assumptions the number rests on
A thirty-year retirement. Retiring at 65 and planning to 95 is thirty years. Retiring at 50 is not.
US market history. The data is one country's returns over one stretch of one century, and it is the century in which that country did unusually well. There is no forward-looking model in the rule. It is a statement about what would have worked, not about what will.
A portfolio held between 50% and 75% in shares, through everything. The rule fails immediately if you sell after a crash. The surviving runs all required the retiree to keep the equity allocation intact during the exact years when doing so felt reckless.
No fees and no taxes. Bengen modelled gross returns. A 1% advice fee comes directly off the withdrawal rate, and money coming out of a traditional 401(k) or IRA is taxable income, so $40,000 withdrawn is not $40,000 spent.
Spending that never changes. Real spending is not flat. It falls in some years, spikes for a roof or a car, and rises again for medical costs late on. The rule assumes none of that happens.
Why the order of returns changes the answer
Two retirees can get exactly the same set of annual returns in a different order and end up with different amounts of money. This is the mechanism that makes the mid-1960s retirees the worst case.
Both start with $1,000,000 and take $40,000 at the beginning of each year. Both get returns of −20%, −10% and +30%. The only difference is the order.
| Year | Bad years first | Good year first |
|---|---|---|
| Start | $1,000,000 | $1,000,000 |
| 1 | −20% to $768,000 | +30% to $1,248,000 |
| 2 | −10% to $655,200 | −10% to $1,087,200 |
| 3 | +30% to $799,760 | −20% to $837,760 |
Same three returns, same three withdrawals, $38,000 apart after three years.
Now take the withdrawals away. Without them both portfolios finish at $936,000, because multiplication does not care about order: 0.8 × 0.9 × 1.3 gives the same product however you arrange it. The entire gap comes from the withdrawals. Selling into a fall converts a paper loss into a permanent one, and there are fewer units left to participate in the recovery.
That is sequence of returns risk, and it is concentrated almost entirely in the first ten years. A bad decade at the end of a retirement barely matters. A bad decade at the start is what the 4% figure was calibrated against. If you want the wider context on what those decades look like, the site covers how bear and bull markets are actually defined and which recession indicators have and have not worked.
Why 4% is too high for a 45-year retirement
Stretch the horizon and the safe rate falls, because there are more years in which a bad sequence can start and more years of inflation compounding against a fixed income.
The drop is not proportional. Going from thirty years to fifty does not halve the rate, because a portfolio that survives thirty years of withdrawals is usually growing by then. The rate converges towards the portfolio's sustainable real return rather than towards zero. Run the same historical data over forty to fifty year horizons and the surviving rate lands near 3.5%, which is why most people planning an early retirement use something between 3.25% and 3.5% rather than 4%.
The difference sounds small and is not. The same $70,000 of annual spending, priced at four rates:
| Withdrawal rate | Multiple of spending | Portfolio needed for $70,000 |
|---|---|---|
| 4.0% | 25.0× | $1,750,000 |
| 3.75% | 26.7× | $1,866,667 |
| 3.5% | 28.6× | $2,000,000 |
| 3.25% | 30.8× | $2,153,846 |
Three quarters of one percentage point is $403,846, or roughly six extra years of saving for most people. It is the reason retiring at 50 is a harder problem than retiring at 65 by more than the fifteen years suggests.
Bengen's own revision
Bengen did not stop in 1994. Adding small-cap stocks to the portfolio raised his figure to 4.5%, and his later work across a broader set of asset classes has put the maximum safe historical rate at 4.7%.
That is not permission to withdraw 4.7%. It is the same exercise on a wider portfolio over the same thirty-year horizon, and it carries every one of the five assumptions above. The revision demonstrates that the number is an output of its inputs. Change the asset mix, the horizon, or the data window, and the answer moves. Anybody quoting a single figure without those three things has quoted half a sentence.
The same rule read backwards
Divide 100 by the withdrawal rate and you get the multiple of annual spending you need saved. At 4% that is 25 times. This is where the FIRE movement's "25x your expenses" comes from, and it is the more useful form, because it turns an abstract rate into a target.
It also makes the arithmetic of spending visible. Every $1,000 of annual spending you remove permanently cuts $25,000 off the target at 4%, and $28,600 at 3.5%. Cutting spending does two things at once: it lowers the number and raises the savings rate that gets you there. Nothing else in the calculation has that property, which is the argument made at length in The Simple Path to Wealth.
The reverse reading has one trap. The 25 times figure is 25 times your retirement spending, not your current spending. A mortgage that finishes in eight years is not part of the number. Commuting costs and work clothes are not part of the number. Health insurance you currently get from an employer very much is.
What to use instead of a fixed rate
The rule's weakness is that it commits to a spending level in year one and never looks at the portfolio again. Every practical alternative fixes that by letting spending respond.
Guardrails. Set a starting rate, then a ceiling and a floor. If the portfolio falls far enough that your current withdrawal represents more than, say, 5% of it, you cut spending by 10%. If it rises far enough that you are below 3%, you give yourself a rise. Most of the time neither trigger fires.
Skip the inflation rise after a down year. Freezing the increase in the year after a loss is a small change that removes a large amount of the damage, because it stops the withdrawal growing at the exact moment the portfolio cannot support it.
Hold a cash buffer. Two to three years of spending in cash or short bonds means the first bad years are funded without selling shares. This does not raise the safe rate much on paper, and it is the difference between a plan you follow and one you abandon in month eight.
Count the income that arrives later. Social Security starting at 67 or 70 permanently reduces what the portfolio has to produce from that point on. A plan that ignores it overstates the target considerably.
Where this fits
The 4% rule is a useful piece of arithmetic and a poor plan. As a way of turning annual spending into a savings target it is the best single number available. As an instruction for what to withdraw in year fourteen of a forty-five year retirement it is being asked to do something it was never tested for.
If you are working out what your own number is, the site has the pieces: the two tests for retiring at 50, a 401(k) projection calculator for what the balance grows to, what people actually have by age, and which accounts to fill first.
Frequently asked questions
Is the 4% rule still valid?
It is still an accurate description of what survived US market history from 1926 to 1992 over thirty-year periods. Whether it is valid for you depends on whether your retirement resembles that test. A thirty-year retirement with a 60% equity portfolio and low fees is close. A forty-five year retirement with a 1% advice fee is not.
Do I take 4% of the balance every year?
No. That is a different strategy. The rule takes 4% once, in year one, and then increases that dollar figure by inflation annually regardless of what the portfolio does. Taking a percentage of the current balance each year cannot run out but produces an income that swings with the market.
What withdrawal rate should I use for early retirement?
Between 3.25% and 3.5% is the range most long-horizon planning uses, which puts the target at 28 to 31 times annual spending rather than 25. The exact figure matters less than knowing which one you used, because it sets the size of the portfolio you are saving towards.
Does the 4% rule account for taxes?
No. Bengen modelled gross portfolio returns and gross withdrawals. Money leaving a traditional 401(k) or IRA is ordinary income, so a $40,000 withdrawal might be $32,000 of spending. Money leaving a Roth account after the five-year rule is satisfied is not taxed, which is why the mix of accounts changes the rate you effectively get.
What is sequence of returns risk?
The risk that poor returns arrive early in retirement rather than late. Because you are selling to fund spending, an early fall permanently removes units from the portfolio that cannot participate in the recovery. The same annual returns in a different order produce a different result once withdrawals start, and no different result before they do.
How much do I need to retire on $50,000 a year?
At 4%, $1,250,000. At 3.5%, $1,428,571. At 3.25%, $1,538,462. Subtract the present value of any pension or Social Security that will cover part of that $50,000, because the portfolio only has to produce the remainder.
Does the 4% rule work outside the United States?
The study used US data, and the US had one of the strongest equity records of the twentieth century. Work using broader international data has generally produced lower safe rates. If your portfolio is globally diversified, which most index investors' portfolios now are, the US-only figure is optimistic rather than conservative.