How much can you spend from a retirement pot each year without running out? The best-known answer is the "4% rule": withdraw about four percent of your starting portfolio in year one, then adjust that amount for inflation each year after. It emerged from historical research and offers something genuinely valuable — a rough anchor for a terrifyingly open question.
It is a useful compass. The mistake is treating a rule of thumb as a guarantee accurate to the decimal.
What it is really saying
The rule’s practical translation is a target: to fund a given annual spend, you need a pot roughly twenty-five times that amount. It exists because spending too aggressively risks depleting savings, while spending too timidly means needlessly sacrificing the life the money was for. Four percent was, historically, a level that survived a wide range of market conditions.
That framing alone is worth the price of admission: it turns "am I saving enough?" into a concrete number you can aim at.
Where it strains
The rule is a simplification, and its assumptions can bend. It was based on particular market histories and a set retirement length; different conditions, a very long retirement, high fees, or the sequence-of-returns problem can all change the safe number. Retirement spending is also rarely a smooth inflation-adjusted line — real life spends unevenly.
None of this makes the rule useless. It makes it a starting estimate to be stress-tested against your own situation, not a promise to lean your whole future on.
Using it wisely
Treat four percent as a planning anchor, then build in flexibility. Being willing to trim withdrawals in bad market years dramatically improves the odds your money lasts. Holding a cash buffer, watching fees, and revisiting the plan periodically all matter more than defending an exact percentage.
The rule’s enduring gift is the twenty-five-times target — a clear finish line to save toward. Just hold the precise number loosely, and let the flexibility around it do the real work.
What the original study actually tested
The rule emerged from research examining every historical thirty-year period in one major market, asking what withdrawal rate, adjusted annually for inflation, would have survived all of them. The answer was close to four percent for a portfolio split between equities and bonds.
Three features of that design matter and are frequently dropped in the retelling. It was a backtest of historical sequences rather than a forecast. It assumed a rigid withdrawal continuing regardless of conditions, which nobody actually does. And it assumed no fees, no taxes and no changes to the portfolio.
None of these invalidate the finding; they define its scope. The study answered a specific and useful question — what would have worked historically under these assumptions — and it did not answer the question people use it for, which is what will work for me.
The assumptions that move the number most
Several inputs shift the sustainable rate substantially, and it is worth knowing which. Fees are the clearest: a percentage taken annually comes directly off the withdrawal capacity, and a portfolio charged one percent is not the portfolio the study modelled.
The horizon is next. Thirty years suits a retirement beginning in the mid-sixties. Forty years does not, and anyone stopping substantially earlier should treat the figure as an upper bound rather than a target. The direction of the adjustment is not small.
The market history used is the third and least discussed. Results drawn from the strongest large market of the twentieth century are systematically more encouraging than results from a broader international sample, and using a single favourable record introduces an optimism nobody intended. Each of these arguments for a lower starting figure than the famous one.
What flexibility buys
The rigid-withdrawal assumption is what makes the rule conservative, and relaxing it is where most of the available improvement sits. A retiree who reduces withdrawals during a poor stretch is meaningfully safer than the fixed figure implies.
Several formalised approaches exist: withdrawing a fixed percentage of the current balance rather than an inflation-adjusted fixed amount, applying guardrails that adjust the withdrawal when the portfolio moves beyond set bounds, or simply deferring discretionary spending during bad years.
Each trades certainty of income for certainty of not running out, which is a reasonable trade for anyone whose spending contains a genuine discretionary component. Someone whose entire projected spending is essential has nothing to flex and needs a lower starting rate to compensate, which is the situation the rigid assumption actually describes.
Where the rule strains hardest
Beyond the assumptions, the rule is silent about several things that materially affect a real retirement. It does not account for spending that changes shape over time, and the evidence suggests it usually does: higher in the early active years, lower in the middle, and potentially much higher at the end if care is needed.
It does not account for other income arriving partway through, such as a state pension starting several years after retirement, which means a higher withdrawal in the early years may be entirely sustainable. It does not account for tax, which differs by the account the money comes from.
And it says nothing about what happens if you are wrong. A plan built on the rule has no defined response to being on a failing path, which is the thing that actually matters, since the response determines whether a poor sequence produces an adjustment or a crisis.
Using it as an orientation rather than a plan
The rule's genuine value is that it converts a formless anxiety into a checkable number using arithmetic anybody can do in a minute. Multiply intended annual spending by twenty-five and you have an order of magnitude, which is enormously more useful than no figure at all.
That makes it a good first instrument. It indicates whether the current saving rate is broadly in the right region, decades out, when precision is impossible anyway and the useful question is directional.
It is a poor final instrument. Within a few years of the decision, the questions become specific — sequencing, tax, the cash runway, the transition — and a single multiplier addresses none of them. Continuing to plan with it at that point is applying a rough tool to a problem that has become precise.
A more honest way to state the answer
The most defensible use is to produce a range rather than a number. Run the calculation at a conservative withdrawal rate with pessimistic assumptions, at the standard figure, and at an optimistic one. The spread between the three is genuinely informative.
A plan that works at the conservative end is robust. One that only works at the optimistic end requires favourable conditions, which is worth knowing while there is still time to change something. That framing also stops the figure being treated as a threshold that has been either met or failed.
Above all, the number should not be quoted with more precision than the method supports. It came from a historical backtest with simplifying assumptions, and treating its output as a target to be hit exactly is asking it to carry weight it was never built for. As with everything on this site, this is educational rather than advice, and anyone close to this decision should consider having it reviewed properly.
What to multiply, which matters more than the multiplier
Most of the error in applying this rule comes from the input rather than the rate. The figure being multiplied should be what your own assets must generate, which is total required spending minus every other reliable income source.
Those sources are usually substantial and frequently omitted. State provision, any defined benefit entitlement, rental income, part-time earnings, and in some households a partner's separate provision. Subtracting them before multiplying can reduce the portfolio target dramatically, and for lower earners state provision alone can cover a large share of essential costs.
The other adjustment is to the spending figure itself. Costs at retirement are not current costs: commuting, work expenses and the contributions being made to the plan all disappear, while healthcare and support costs tend to rise later. Building the figure from components rather than estimating a total is an evening of work and it changes the answer more than any refinement to the withdrawal rate.
The other direction of failure
The rule is designed to answer the question of whether money will run out, and the historical testing found that in most sequences it did not merely survive — it finished with a balance considerably larger than it started with.
That is worth stating because it identifies a second kind of failure that receives almost no attention. Someone who applied a deliberately conservative rate, lived well below what their assets supported for thirty years, and died with a very large balance did not have a successful retirement; they had an unnecessarily constrained one.
The response is not to withdraw recklessly but to build in a mechanism for adjusting upward when the portfolio is clearly ahead, in the same way flexible approaches adjust downward when it is behind. A rule that only ever moves in one direction guarantees one of the two failures, and the underspending one is the more common outcome.
Why a single figure travels so well
It is worth noticing why this rule, of all the retirement research produced over several decades, is the one everybody knows. It is not the most sophisticated, the most current, or the most applicable outside the market it was derived from.
It spread because it is a single number, applied by multiplication, producing an answer to a question people find frightening and otherwise unanswerable. That combination is what makes an idea travel, and it is the same set of properties that made the budgeting rule discussed elsewhere on this site so widely repeated.
The hazard is identical too: a heuristic optimised for memorability gets treated as though it were optimised for accuracy. Holding both facts at once — that it is a genuinely useful starting point and that its precision is illusory — is the right way to carry it, and it is more or less the only defensible way. None of this is financial advice.
What to do when the calculation looks impossible
A great many people run this for the first time, arrive at a figure that seems unreachable, and stop engaging with the question entirely. That is the worst available response, because every lever that would improve the position works better with more time.
The first move is to check the inputs, since the initial calculation almost always overstates the requirement by using current spending unadjusted and ignoring state and workplace provision. Correcting both frequently reduces the target substantially, and occasionally enough to change the assessment completely.
The second is to recognise that the target is not binary. Assets covering half your requirement means working part-time rather than full-time, or stopping later rather than never, or having a floor under a difficult period. The framing in which anything short of the number is failure is inaccurate, and it is the single most common reason people abandon the exercise entirely.