Multiply by 25, or by 33, or by 20
A 4 percent withdrawal rate means the pot is 25 times the annual withdrawal, because 1 divided by 0.04 is 25. At 3 percent the multiple is 33.3, and at 5 percent it is 20. Those three numbers are the whole page. What makes the choice consequential is that the multiple does not move gently: dropping the rate by a single percentage point, from 4 to 3, raises the required assets by a third.
The other lever is the income that already arrives. Subtracting a pension before applying the multiple, rather than after, is what makes the requirement collapse for people who have one. A $2,000 monthly pension removes $600,000 from a 4 percent target, and $800,000 from a 3 percent one. It is the largest single item on most people's page and it is often left out of casual retirement arithmetic entirely.
Where the 4 percent figure came from
William Bengen published the original analysis in 1994, testing withdrawal rates against historical US market returns and asking which rate survived every rolling 30-year period, including retirements that began just before the worst of them. Around 4 percent did, with the withdrawal raised each year for inflation. The Trinity study a few years later reached similar conclusions with a different method, and the number entered general circulation as a rule.
Three caveats travel less well than the number does. It was derived from US market history, which was unusually good by international standards, and studies using other countries' data generally produce lower rates. It assumed a 30-year retirement, so someone stopping at 55 is outside the tested range. And it assumed a specific mix of stocks and bonds held throughout, with fees ignored, which means a portfolio charging a percent a year is not the portfolio that was tested.
Sequence risk is the reason a fixed rate is only a starting point
Two portfolios with identical average returns can leave a retiree in completely different positions depending on the order the returns arrived in. Someone drawing an income sells units to fund it, and selling during a fall means selling more units at a lower price, permanently reducing what is left to recover. The same bad years arriving late in retirement, after two decades of growth, barely register.
This is why a fixed percentage is a planning starting point rather than an operating rule. The common adjustments are all forms of flexibility: reducing withdrawals in years after a fall, skipping the inflation increase after a bad year, or holding one to two years of spending in cash so the portfolio does not have to be sold at the worst moment. None of these are modelled here, and all of them make a given rate more survivable than a rigid version of it.
How to use the number this page gives you
Treat the result as the scale of the problem rather than the specification of a plan. A figure of $900,000 tells you whether you are looking at a decade of saving or three, and whether the gap is closed mainly by saving more, working longer, or spending less in retirement. Those are the three levers and they trade against each other directly. What the figure will not do is tell you that $900,000 is enough, because that depends on how long you live, what returns arrive and in what order, and what happens to your costs, and none of those are known now.
For a more complete version that includes inflation between now and retirement, a defined end date and a required monthly saving, use the retirement savings calculator. To model the growth of the pot on its way there, the savings calculator works from a monthly contribution and a target.
Questions people ask
Is 4 percent safe?
It was safe in the historical US data Bengen tested, which is a narrower claim than it is usually repeated as. Applied to other countries' return histories the equivalent figure is generally lower, and applied to retirements longer than 30 years it is lower again. Fees were not deducted in the original work, so a portfolio charging one percent a year is effectively testing a higher withdrawal rate than the label suggests. Treat 4 percent as a well-studied reference point that came out of a specific dataset, not a property of markets.
Should I include a pension in the pot or subtract it from the income?
Subtract it from the income, which is what this page does. A pension is a stream, not a balance, and converting it into an implied lump sum then adding it to your assets produces a figure you cannot withdraw from and that behaves nothing like invested money. Subtracting it first also makes the sensitivity clear: the pot only has to cover the gap, so a larger pension shrinks the requirement much faster than the same amount of extra assets would.
Does this account for inflation?
Yes, in the sense that everything is in today's money and the withdrawal rate convention assumes you raise the dollar amount with inflation every year, which is how the original studies defined it. What it does not do is translate the target into the dollars of a future year, which is a separate step and matters a great deal if retirement is decades away. The retirement savings calculator on this site does that translation and shows both figures side by side.
What if I retire early?
The rate that was tested for 30 years is not the rate for 45. Longer horizons need lower withdrawal rates, and the relationship is not linear, since a very long retirement is exposed to more of the sequences that break a plan. People planning to stop well before conventional retirement age generally use 3 to 3.5 percent, add flexibility to cut spending after bad years, or plan on some income continuing. This page lets you set the rate directly for that reason rather than presenting one figure as the answer.
What happens if returns are bad in the first few years?
That is the scenario a fixed withdrawal rate handles worst. Selling to fund an income during a fall removes more units than the same withdrawal would in a rising market, and those units are not there to recover, so a poor first decade can leave a portfolio permanently behind one that saw the identical returns in the opposite order. The practical responses are to hold a cash buffer so the portfolio is not sold at the bottom, and to reduce or freeze withdrawals for a period after a serious fall. Neither is modelled here, and both make a given rate substantially more robust.