How Long Will $1 Million Last in Early Retirement? The Historical Answer

Straight-line math says $1M at $40k a year lasts 25 years. Real US market history says the median window lasted the full 40-year test - and the worst ran dry in 16.6 years. Here's how long $1 million actually lasted at $40k, $60k, $80k, and $100k of annual spending.

Straight-line math says $1 million at $40,000 a year lasts exactly 25 years. That number is almost never what happened. Across real US market history since 1871, a $1M all-stock portfolio spending $40,000 a year (adjusted for inflation) more often lasted the full 40 years we tested, and in the median case still had money at the end. The reason is simple: the balance keeps growing while you spend it, so the money outruns the arithmetic.

But the same history hides a sharp edge. The worst starting point on record, October 1929, ran that same $40,000 plan dry in 16.6 years, well short of even the naive 25-year figure. That gap between the typical outcome and the worst one is the whole story of retirement risk, and it is invisible if you only ever divide your savings by your spending.

How we measured it

We took $1,000,000 in real (inflation-adjusted) dollars and walked it forward through every possible starting month in the bundled US stock series (Shiller, real total return with dividends reinvested, 1871 to 2025). Each month the balance grows by that month’s actual real return, then a fixed real withdrawal of one-twelfth of the annual spend comes out. When the balance hits zero, we record how many years it lasted. We ran a full 40-year test on every one of the 1,387 starting months that has 40 years of data behind it, so every window is a complete, comparable outcome.

This is a straight, undisturbed historical backtest of a single all-US-stock portfolio. It is deliberately simpler than the simulator itself, which models US, international, and bond allocations and resamples history in blocks. Read the numbers below as an honest illustration of sequence risk on an aggressive portfolio, not as a forecast and not as the tool’s multi-asset math.

The headline: history beat the straight-line math

Here is how long $1 million actually lasted at four common spending levels, next to the straight-line answer everyone reaches for first.

Illustrative only, not advice. $1,000,000 in real dollars, 100% US stocks (Shiller real total return), across 1,387 historical 40-year windows since 1871. "Median history" is the middle outcome; "40+ yr" means the median window still had money at the 40-year test cap. Straight-line = $1M divided by annual spend. Historical backtest of real starting points, not a forecast.

At the low end the two columns barely relate to each other. Straight-line, $60,000 a year drains $1M in 16.7 years. In real history, the median $60,000 window lasted the full 40 years, and better than two in three starting points made it past 30. Growth did the heavy lifting: a portfolio compounding at even a modest real rate refills the account faster than a 4-6% withdrawal empties it, most of the time.

The phrase “most of the time” is where the higher spend levels bite. At $80,000 a year the median window lasted 24.3 years, nearly double the straight-line 12.5, but only 39.7% of starting points reached 30 years. At $100,000 a year the median and the straight-line answer nearly converge, because at a 10% withdrawal rate you are spending faster than a stock portfolio can reliably grow, and fewer than one in five windows survived three decades.

The downside: what an unlucky start looked like

An average is a comforting number and a dangerous one to plan around, because you only retire once and you do not get to average across a hundred timelines. So here is the other half of the picture: the unlucky-but-not-worst case (the 10th percentile, meaning one in ten starts did at least this badly) and the single worst starting month in the entire record.

Illustrative only, not advice. Same 1,387 windows as above. "10th percentile" means one in ten historical starts lasted no longer than this; "40+ yr" means that outcome still had money at the 40-year cap. Every spend level's worst window began in October 1929. Historical backtest, not a forecast.

Two things jump out. First, even at a cautious $40,000 a year, the worst-case history was harsh: a retiree who started in October 1929 saw $1M last 16.6 years, not the 25 the math promised and nowhere near the 40+ the median enjoyed. Second, every single worst case in the table shares one start date. That is not a coincidence, it is the signature of sequence-of-returns risk. Retiring straight into the 1929 crash meant selling shares into a collapse to fund living expenses, permanently shrinking the base that later recoveries had to rebuild from. The same crash, arriving ten years into retirement instead of on day one, would have done a fraction of the damage. Sequence risk is the reason the order of returns matters as much as the average, and it is exactly what a straight-line number cannot show you.

So how long will your $1 million last?

The honest answer is a range, not a number, and its width is set almost entirely by your spending rate. Spend around 4% of it a year and history says $1M was very likely to outlast a normal retirement, with a comfortable margin even in bad starts. Push toward 8-10% and you move into territory where the typical outcome is fine but a meaningful share of real histories ran out, and the unlucky ones ran out fast. This maps directly onto the 25x rule and safe withdrawal research: $1M supports roughly $40,000 of spending at the 4% guideline precisely because that is the rate history was kind to.

If you are still building toward the number rather than spending it, the same lesson applies in reverse: the size of the portfolio matters, but so does not being forced to sell into a downturn in the first few years. That is the logic behind coasting and part-time income bridges, which cut early withdrawals exactly when they do the most harm.

One caveat governs all of it. This is the record of US markets, which delivered an unusually strong century and a half, and the future is not required to repeat it. A backtest tells you which starting points would have worked, not which ones will. Treat these figures as a well-documented map of the past, useful for seeing the shape of the risk, and let your own plan lean a little more cautious than the luckiest version of history.

Other starting amounts: $500,000 and $2 million.

Frequently asked