How Long Will My Money Last in Retirement?
Quick answer: There is no single number of years, because the answer has two parts. One part is plain arithmetic: a balance divided by yearly spending gives the years it would cover if markets did nothing at all. The other part is how the markets behave while you draw on the money, which a simulation can only sketch as a spread of outcomes. Under the calculator's default assumptions (1971 to 2025 real returns, a 60/40 mix of stocks and bonds, no other income), a $1,000,000 start with $40,000 of yearly spending, offered here as an example, lasted to age 95 in 95% of the simulated paths and to age 105 in 87%. Change the balance, the spending or the age you plan to, and those numbers move a lot.
Education, not financial advice. This page explains a method and shows what a model produced from stated assumptions. It does not tell you what to buy, sell, save or spend. For your own situation, ask a qualified professional.
The question hides two different questions
When people ask how long their savings will last, they usually mean one of two things. The first is a counting question: if I spend this much each year, how many years of spending does this balance hold? The second is a risk question: across the many ways the next few decades could unfold, how often does the balance hold out until the age I care about?
The counting question has one answer. The risk question has a spread of answers, and this page is mostly about that spread. A Monte Carlo tool answers the second question by running thousands of imagined retirements, each with its own sequence of yearly returns, and counting how many still have money left at the end. If you have not seen how that works, what a Monte Carlo simulation is is explained in plain terms on its own page, and you can run your own numbers in the Monte Carlo simulation retirement calculator on the home page.
Everything below is in today's dollars. The model uses real returns, meaning returns after inflation, so, as an example, a $40,000 withdrawal in year twenty is assumed to buy what $40,000 buys now. That keeps the comparisons simple, and it also means the model's inflation is the historical average baked into the dataset, not a rate you choose.
Start with the arithmetic that needs no model
Before any simulation, a pencil gives you a floor on the question. Divide the balance by the yearly spending and you get the number of years it would last if the money earned nothing and prices never moved. As an example, $500,000 divided by $30,000 a year is about 16 years. The same $500,000 at $40,000 a year is about 12 years, at $50,000 about 10 years, and at $60,000 about 8 years. Start with $1,000,000 and $40,000 a year, and the division gives 25 years. Start with $1,500,000 and $60,000 a year, and it also gives 25.
Those figures are not predictions. They are the answer to a deliberately silly question, namely what happens with zero returns. Real portfolios earn something in most years and lose something in others, so the true outcome sits above that floor in many paths and below it in the paths where early losses arrive before the withdrawals have slowed. What the simulation adds is a sense of how wide that gap is.
Twelve worked cases from the simulation
The table below runs twelve starting points through the calculator's engine. Each case starts at 65 with the stated balance, draws the stated yearly amount at the start of every year, and uses a 60/40 mix with the default 1971 to 2025 real-return assumptions. There is no Social Security or pension in these cases, so every dollar of spending comes from the portfolio. The success column is the share of simulated paths that still had money left at age 95. The age column comes from a separate run to age 100, and it shows the middle age at which the money ran out among only those paths that did run out.
| Saved at 65 | $30,000 a year | $40,000 a year | $50,000 a year | $60,000 a year |
|---|---|---|---|---|
| $500,000 | 59%ran out near age 88 | 19%ran out near age 82 | 3%ran out near age 78 | 0%ran out near age 75 |
| $1,000,000 | 99%ran out near age 95 | 95%ran out near age 93 | 80%ran out near age 91 | 59%ran out near age 88 |
| $1,500,000 | 100%ran out near age 99 | 100%ran out near age 96 | 99%ran out near age 95 | 95%ran out near age 93 |
A few readings stand out in this example set. Take the $500,000 row at $30,000 a year. The 30-year success rate is 59%, and the paths that ran dry did so at a middle age of 88. Raise the spending in that same row to $40,000 and the success rate falls to 19%, with the failing paths running dry at a middle age of 82. At $60,000 a year the same balance reaches age 95 in 0% of paths, which is a way of saying that it essentially never does.
In the example rows, the $1,000,000 start is where small changes in spending matter most. At $30,000 a year the success rate is 99%. At $40,000 it is 95%. At $50,000 it is 80%, and at $60,000 it is 59%. Each extra $10,000 of yearly spending cost this balance more than the one before it, because every larger withdrawal also removes the future growth the withdrawn dollars would have earned.
For the $1,500,000 example the table is nearly flat for lighter spending: 100% at $30,000 and 100% at $40,000, with 99% at $50,000 and 95% at $60,000. Even there, the last row is not a round 100, and the few paths that fail run out at a middle age of 93, late in the run.
What this number leaves out. It comes from a model with the assumptions shown, in annual steps and after-inflation returns. It does not model taxes, health or long-term care costs, inflation that differs from the model, or how you would act in a bad year. See what a Monte Carlo simulation cannot see.
Why the middle failing age sits later than the arithmetic
Compare two numbers for the example $500,000 row at $30,000 a year. The zero-return arithmetic says about 16 years, which would carry the money to roughly age 81. Yet the paths that failed in the simulation failed at a middle age of 88. Is the model contradicting the arithmetic?
It is not. Failing paths are the unlucky ones, but unlucky relative to the average, not relative to zero. A path that earns something small in most years still stretches the balance beyond the no-returns floor; it just falls short of the longer stretch you were hoping for. The spread of ages matters here too. The figure in the table is a middle value, so half of the failing paths ran dry earlier than it and half later.
This also explains why a failure age should never be read as "the age I will run out". It describes what happened in the unlucky share of imagined paths. If you are in the lucky majority, there is no such age at all.
How the age you plan for changes the answer
The second table keeps the balance and the spending fixed and moves only the planning horizon. The example is the same $1,000,000 starting at 65 with $40,000 of yearly spending, 60/40, default assumptions, and no other income. Each row asks for the money to last until a different age.
| Lasts to age | 80 | 85 | 90 | 95 | 100 | 105 |
|---|---|---|---|---|---|---|
| Share of paths | 100% | 100% | 98% | 95% | 91% | 87% |
Reading across, the success rate is 100% to age 80, 100% to age 85 and 98% to age 90. Pushing the horizon to 95 gives 95%, to 100 gives 91%, and to 105 gives 87%. The slope is gentle at first and steeper later, because each added year both adds a withdrawal and lets one more stretch of market history play out.
Notice what this table does not say. It does not say how likely you are to live to any of those ages. That is a separate question about people, not portfolios, and this page prints no life-expectancy figure. The figures we would need come from the Social Security Administration's period life table, which we could not read when we prepared this page, so we have left them out rather than quote them from memory. If you want a sense of how long a retirement might need to run, you can look up that table yourself and then use the planning-age field in the calculator to see the matching row.
Many people also notice that the planning age is a choice about risk tolerance, not a forecast. Setting it to 105 does not mean expecting to reach 105; it means asking what the model produced for a very long retirement. Whether a given horizon fits your own situation is a question for you and, if you have one, a professional who knows your circumstances.
What steady outside income changes
All twelve cases above ask the portfolio to cover every dollar. Most real retirements have something else: Social Security, a pension, an annuity, rental income. The calculator models a simple version of this with a yearly amount that begins at an age you choose and does not depend on the markets. That amount reduces the withdrawal from the portfolio each year it applies.
The default plan shows how much it can matter. That plan is age 45, retiring at 65, planning to 95; $300,000 saved, $10,000 saved a year, $65,000 a year of spending in today's dollars, $24,000 a year of Social Security from 67, 60% stocks and 40% bonds, and it produced 87% success under the default assumptions. Remove the Social Security line and leave everything else untouched, and the figure falls to 64%. Same savings, same spending, same markets; the only difference is whether a fixed stream of income covers part of the bill from age 67. In the model, steady income does two things at once. It lowers the number of dollars the portfolio must supply, and it lowers them in exactly the years when selling investments after a market drop is most costly.
That connects this page to a bigger idea. Readers who want the full mechanics will find it under sequence of returns risk, where the same set of yearly returns is shown in two different orders. The short version: the earlier a drawdown arrives relative to your withdrawals, the more of its damage is permanent.
A look at one historical start
The simulation draws random sequences, but history only produced one. In the dataset we use, a 60/40 mix with real returns from Damodaran's file, a 30-year retirement that began in 1966 and withdrew an example $40,000 a year from $1,000,000 ran out of money in 1991. The 1982 start, on the same terms, never ran out and ended with $5,360,000 in today's dollars. Among the 69 thirty-year windows the dataset allows, 65 of 69 lasted the full 30 years; the starts that did not were 1965, 1966, 1968, 1969.
Two cautions go with those figures. The windows overlap, so they are not 69 independent experiments, and a handful of long bad stretches drive the whole result. Second, they come from one dataset of annual data; they show what that history did, not what any future start will do. The rule of thumb behind the $40,000-on-$1,000,000 example has its own page, under the name 4% rule, and the broader menu of ways to take money out is covered in withdrawal strategies.
What moves the answer
If you run a few experiments in the calculator, a pattern appears. Spending enters every year of the run, so changes to it echo through everything, as the twelve cases show. The planning age matters through the number of extra years it adds. The stock and bond mix matters too, though in this model less dramatically than the other inputs: for the $40,000-on-$1,000,000 example run over 30 years, the most bond-heavy mix tested scored 89% and the 60/40 mix scored 95%. The return assumptions can move a score by more than that. Lowering the stock return assumption by one point took the default plan from 87% to 80%. For help with what the percentage itself means, see what a success rate means, including why a score near the top is not necessarily the target.
Two other things shape the experience of a long retirement without appearing in the table at all. One is flexibility: a person who can spend somewhat less after a bad year is playing a different game from one who cannot, and the simulation in its standard form assumes the fixed amount is withdrawn regardless. The other is structure. Some retirees divide their money into time-based compartments, the idea behind a bucket strategy, and the page on it tests what that does and does not change in the model.
Finally, some readers wonder whether adding a different kind of asset would alter the picture. The page on gold in a retirement portfolio shows what gold did in dated periods of the dataset and runs the calculator both ways, with the dataset's gold return and with gold earning nothing after inflation, because the answer depends heavily on that assumption.
What this can't tell you
This page can't tell you how long your money will last. It does not know your real spending, how your costs will change, what you will owe in tax, whether you will need long-term care, how long you will live, or what the markets will do. It works with a single set of historical return statistics, annual steps and a fixed spending line. The twelve cases and the longevity rows are examples chosen so that the shape of the relationship is easy to see, and each is a statement about simulated paths under stated assumptions, not about yours. For the longer list of what sits outside the model, see what a Monte Carlo simulation cannot see.
If you want to see how the engine reaches these results, the methodology page walks through the order of cash flows and the assumptions. And if you would rather test your own balance, spending and age, the retirement Monte Carlo calculator on the home page runs in your browser and sends nothing anywhere.
Frequently asked questions
Is there a rule for how many years a balance will last?
Only the arithmetic one: divide the balance by yearly spending to see how many years it covers with no returns at all. Simulations then show how that floor shifts once markets are included. Neither the arithmetic nor the simulation is a promise, and each depends on spending staying roughly where you set it.
Why does a small change in spending make such a large difference?
Because each extra dollar withdrawn is a dollar that no longer grows, in every later year. In the twelve cases above, moving from lighter to heavier spending on the same balance changes the simulated outcome by far more than the withdrawal alone would suggest, and the effect compounds over decades.
Does the calculator tell me how long I will live?
No. It takes a planning age that you choose and shows what the model produced for it. It has no life-expectancy data, and this page prints none either. Official period life tables exist for that purpose and can be looked up separately from the calculator.
Do the results here match my own run?
They will match only if your inputs and assumptions match the stated ones. The twelve cases start at 65, use no other income, and run on 1971 to 2025 real returns with a 60/40 mix. Different inputs give different results, and a different random seed can shift a score by a point or two; the default plan ranged 87% to 89% across eight seeds.