One study, thirty years, one country
In 1998 three professors at Trinity University ran a simple test: take a portfolio, withdraw a fixed inflation-adjusted amount each year, and count how often it survived 30 years of real US market history. Around 4% of the starting balance came out as the highest rate that almost never failed, and the authors' own account of the work remains better reading than any summary of it.
Nearly everything load-bearing in that description gets dropped when it hardens into "the 4% rule". The test ran over 30 years, where a retirement at 45 implies 45–55 of them. It withdrew a fixed real amount and ignored markets entirely, no cutting back in a crash, no spending more in a boom. And it used US returns, historically among the strongest of any market. "Never failed historically" is a smaller promise than it sounds, too: it means the rate survived a few dozen overlapping thirty-year sequences from one country's past, not that it carries a guaranteed probability into the future.
None of that makes 4% wrong. It makes it a starting point that has to be adjusted for a horizon and a country the original study never covered. For a 30-year retirement, 4% remains defensible; for 40+ years most researchers land between 3.25% and 3.5%. Where you sit in that band depends on flexibility, because a withdrawal rate is not a plan, it is a starting assumption, and real retirees adjust. A strategy that trims spending in bad years survives rates that would sink a rigid one, which is what earns you the higher end; if every pound of spending is committed, take the lower. Withdrawal strategies compares the flexible approaches side by side, and the history behind the figure (where Bengen got 4%, the conditions attached to it, and the four situations where quoting it plainly goes wrong) is set out in the 4% rule explained.
The curve between 5% and 2.5%
The arithmetic is unforgiving. Your target pot is annual spending divided by the rate, so lowering the rate raises the target on a curve, not a line:
| Rate | Multiple of spending | Pot for £32,000/yr |
|---|---|---|
| 5.0% | 20× | £640,000 |
| 4.0% | 25× | £800,000 |
| 3.5% | 28.6× | £914,000 |
| 3.0% | 33.3× | £1,067,000 |
| 2.5% | 40× | £1,280,000 |
Dropping from 4% to 3% adds £267,000 to the target for the same lifestyle. For most people that is several extra years of work, which is why the rate deserves more thought than any other single input on this site. Fees move you along the same curve from the other side: a 1% platform-and-fund drag effectively turns a 4% withdrawal into 5% of gross return needed, and the fee calculator shows what that costs over a retirement.
The 25× shorthand lives on this curve as well. It is the same statement inverted, since 25× spending is exactly what a 4% withdrawal implies; the 25× calculator is the quick version, and this page is for interrogating the rate itself.
Marcus, six years wide
Marcus is 52 with £820,000 invested and £34,000 a year of spending. At 4% he needs £850,000; he is essentially there, about a year away. At 3.5% the target is £971,000, roughly four more years. At 3.25% it is £1,046,000, closer to six. Same portfolio, same lifestyle, and a six-year spread resting purely on how cautious the rate is. The tool above runs his plan against both simulated and historical outcomes at each rate, so the decision becomes a trade he can see: retire sooner with a thinner margin, or later with a thicker one.
Choosing where to sit in that spread is the practical question, and the walkthrough of this explorer covers how, including why a figure drawn from US history may not travel to a UK portfolio.
Horizon is what stretches that spread. For a 30-year retirement the difference in failure rates between 4% and 3.5% is small. Push the horizon to 45 years and it widens sharply, because the longer the retirement, the more the early years matter, and the more a lower rate buys.
What the rate quietly assumes
Asset mix, for one. Holding more equities let higher rates survive historically, but only up to a point: very bond-heavy portfolios actually failed more often over long horizons because they lost to inflation, and beyond roughly 75% equities the added volatility stops paying for itself.
The rate also stops being fixed the moment you retire, whatever the spreadsheet says. Taking a fixed real amount from a portfolio that has since grown means your current withdrawal rate has fallen, and vice versa; guardrail strategies formalise exactly that observation. Guaranteed income shifts the ground even more. A state or defined-benefit pension starting later means the portfolio only bridges a gap, and can often support a higher early rate than any rule of thumb suggests.
No tax appears anywhere in this arithmetic, and neither does mortality. Planning to 95 when ONS life expectancy figures put the odds of reaching it at modest builds in a margin you may never need, and Rich, Broke or Dead makes that particular trade visible in a way no failure rate can.