The 4% rule says you can spend 4% of your starting portfolio in the first year of retirement, raise that amount with inflation every year after, and not run out over thirty years. Does it hold up? In our data, the 4% rule with 60% stocks lasted 30 years in 90% of 10,000 simulated retirements and in 65 of 69 real periods since 1928. Both numbers matter, and they don't measure the same thing, so the rest of this page shows how they work and where the rule breaks.
Everything below is computed from real US market history for a portfolio of stocks and bonds. Nothing in the tables is typed by hand.
How to read the tables
Each cell has two numbers separated by a bar.
- The first is a Monte Carlo success rate. We simulate 10,000 retirements by drawing years at random from the historical record. It is the share of those retirements where the money lasted.
- The second is a count of real periods. We replay every actual retirement start year in the data (1928 to 2025) that has enough years ahead of it. "N of M" means the money lasted in N of the M real periods.
The plan behind every cell is the same. You start with $1,000,000. The first year you withdraw the chosen rate times that balance, then raise the withdrawal with inflation each year. Money lasts if the balance never reaches zero before the last year. Portfolios are shown as a stock and bond split, so sixty percent stocks and forty percent bonds is the classic mix.
The two numbers can differ. The simulation can produce sequences of years that never happened in a row. The real periods are the only sequences history actually gave us, but they overlap heavily, so they are far fewer independent tests than the count suggests.
Success rate over thirty years
Thirty years is the length the 4% rule is usually stated for. Read across a row to see what a higher withdrawal does. Read down a column to see what the stock share changes.
| Portfolio | 3% | 3.5% | 4% | 4.5% | 5% |
|---|---|---|---|---|---|
| 100% stocks / 0% bonds | 95% | 69 of 69 | 91% | 68 of 69 | 86% | 64 of 69 | 82% | 62 of 69 | 76% | 54 of 69 |
| 80% stocks / 20% bonds | 97% | 69 of 69 | 94% | 69 of 69 | 89% | 66 of 69 | 83% | 60 of 69 | 77% | 54 of 69 |
| 60% stocks / 40% bonds | 98% | 69 of 69 | 95% | 69 of 69 | 90% | 65 of 69 | 84% | 58 of 69 | 75% | 50 of 69 |
| 40% stocks / 60% bonds | 99% | 69 of 69 | 96% | 69 of 69 | 89% | 62 of 69 | 79% | 49 of 69 | 66% | 41 of 69 |
A shorter retirement
If you retire later, you need the money for fewer years. Compare this table with the one above: the same rates last more often.
| Portfolio | 3% | 3.5% | 4% | 4.5% | 5% |
|---|---|---|---|---|---|
| 100% stocks / 0% bonds | 99% | 79 of 79 | 97% | 79 of 79 | 95% | 79 of 79 | 92% | 76 of 79 | 88% | 71 of 79 |
| 80% stocks / 20% bonds | 99%+ | 79 of 79 | 99% | 79 of 79 | 97% | 79 of 79 | 95% | 79 of 79 | 91% | 74 of 79 |
| 60% stocks / 40% bonds | 99%+ | 79 of 79 | 99%+ | 79 of 79 | 99% | 79 of 79 | 97% | 79 of 79 | 93% | 74 of 79 |
| 40% stocks / 60% bonds | 99%+ | 79 of 79 | 99%+ | 79 of 79 | 99% | 79 of 79 | 97% | 79 of 79 | 94% | 72 of 79 |
A longer retirement
Retiring early, as in FIRE plans, means the money must last longer. The same rates fail more often here.
| Portfolio | 3% | 3.5% | 4% | 4.5% | 5% |
|---|---|---|---|---|---|
| 100% stocks / 0% bonds | 92% | 59 of 59 | 87% | 58 of 59 | 82% | 52 of 59 | 77% | 46 of 59 | 70% | 41 of 59 |
| 80% stocks / 20% bonds | 94% | 59 of 59 | 89% | 59 of 59 | 83% | 53 of 59 | 77% | 46 of 59 | 69% | 40 of 59 |
| 60% stocks / 40% bonds | 95% | 59 of 59 | 90% | 58 of 59 | 83% | 50 of 59 | 73% | 40 of 59 | 62% | 32 of 59 |
| 40% stocks / 60% bonds | 94% | 59 of 59 | 87% | 53 of 59 | 75% | 38 of 59 | 61% | 27 of 59 | 46% | 16 of 59 |
Which start years failed
Averages hide when the failures happen. The table lists the real start years in which the money ran out, for the thirty-year plan, at three withdrawal rates.
| Portfolio | 4% rate | 4.5% rate | 5% rate |
|---|---|---|---|
| 100% stocks / 0% bonds | 1929, 1930, 1966, 1968, 1969 (5 of 69) | 1929, 1930, 1965, 1966, 1968, 1969, 1973 (7 of 69) | 1928, 1929, 1930, 1937, 1962, 1964, 1965, 1966, 1967, 1968 and 5 more (15 of 69) |
| 80% stocks / 20% bonds | 1966, 1968, 1969 (3 of 69) | 1929, 1962, 1964, 1965, 1966, 1967, 1968, 1969, 1973 (9 of 69) | 1929, 1930, 1937, 1962, 1963, 1964, 1965, 1966, 1967, 1968 and 5 more (15 of 69) |
| 60% stocks / 40% bonds | 1965, 1966, 1968, 1969 (4 of 69) | 1937, 1962, 1963, 1964, 1965, 1966, 1967, 1968, 1969, 1972 and 1 more (11 of 69) | 1929, 1930, 1937, 1956, 1959, 1960, 1961, 1962, 1963, 1964 and 9 more (19 of 69) |
| 40% stocks / 60% bonds | 1962, 1964, 1965, 1966, 1967, 1968, 1969 (7 of 69) | 1936, 1937, 1939, 1940, 1956, 1959, 1960, 1961, 1962, 1963 and 10 more (20 of 69) | 1929, 1936, 1937, 1938, 1939, 1940, 1941, 1946, 1955, 1956 and 18 more (28 of 69) |
Look at which years repeat from row to row. Failures are not spread evenly across history. They bunch up around a few bad stretches, because what hurts a retiree is poor returns and high inflation early on, when the balance is large and withdrawals are being taken out. The same total return arriving later does far less damage. That is sequence of returns risk, and the 4% rule debate covers it in more detail.
The highest rate that survived every real period
Instead of picking a rate and counting failures, this table works backwards. For each portfolio and length it finds the highest withdrawal rate, in steps of a tenth of a percentage point, that lasted in every real period in the data. The year in brackets is the start year that fails first one step higher.
| Portfolio | 20 years | 30 years | 40 years |
|---|---|---|---|
| 100% stocks / 0% bonds | 4.1% (1929) | 3.4% (1929) | 3.3% (1929) |
| 80% stocks / 20% bonds | 4.5% (1969) | 3.8% (1966) | 3.5% (1966) |
| 60% stocks / 40% bonds | 4.6% (1966) | 3.7% (1966) | 3.4% (1966) |
| 40% stocks / 60% bonds | 4.5% (1966) | 3.5% (1966) | 3.2% (1966) |
For a thirty-year retirement with sixty percent stocks, that rate is 3.7% (1966 start). This is the figure that gets quoted as "the safe withdrawal rate", and it comes with a catch: it is set by a single worst period. Add one more bad decade to the data and it moves. A rate that survived every past period was not safe in any guaranteed sense, it was the rate that history happened to allow.
Method and limits
- US data only. The returns, bond yields and inflation come from one country's record, which is one history among several possible ones. Other countries' markets have had different records.
- Annual steps. Returns and withdrawals are applied once a year. Real retirements have monthly cash flow and market moves inside the year.
- Independent years in the simulation. The Monte Carlo draws each year separately, so it ignores any pattern from one year to the next.
- Overlapping real periods. Retirements starting one year apart share almost all their years. 69 thirty-year periods are not 69 independent samples.
- Not included. No taxes, no fund fees, no Social Security or pension income, and no change in spending when markets fall. Your own plan will have some of these.
- Fixed rule. The withdrawal rises with inflation no matter what the portfolio does. People who cut spending after bad years do better than the tables show.
Past results are not a promise. A rate that lasted in the past can fail in the future, and a rate that failed can last. How we build the numbers is on the methodology page.
Test your own plan
The tables use one starting balance and one fixed rule. Your plan has your own savings, spending, ages and mix. Run it through the Monte Carlo retirement calculator to see how often it lasts. If you are still saving and want to know when you could stop contributing, try the Coast FIRE calculator. To see how stocks have behaved over long periods, read how to invest in the S&P 500.
Past returns don't guarantee future results. This article is educational and is not financial advice.
Source: Damodaran, NYU Stern (S&P 500 total return, bonds, gold, T-bills, CPI), 1928–2025.
