The 4% Rule: How Much Do You Need to Retire? (Tested Since 1928)

The 4% rule is the best-known answer to the biggest money question most people ever face. How much do I need to stop working?

It goes like this. In your first year of retirement, take 4% of your savings. Every year after that, take the same amount plus inflation. Do that, and the money should last at least 30 years.

Turn it around and you get your retirement number. 4% is one twenty-fifth, so you need 25 times what you spend in a year. Spend $40,000 and you need $1,000,000.

What the 4% rule actually says

The rule comes from a 1994 paper by William Bengen, a financial planner in California. He took a portfolio of 50% US stocks and 50% medium-term US Treasury notes. Then he checked how much a retiree could have taken out in every historical period without running out.

His answer was about 4% of the starting balance, raised each year by inflation. Here's how that works on $1,000,000:

• Year one: you take $40,000.
• Prices rise 3% that year. Year two, you take $41,200.
• Prices rise 3% again. Year three, you take $42,436.
• You keep going like this, whatever the market does.

Notice what the rule doesn't say. It isn't "take 4% of whatever's left each year". Your income stays steady in real terms, and the portfolio takes the hit when markets fall. That's what makes it useful, and it's also what can make it fail.

How much do you need to retire?

Divide your yearly spending by the withdrawal rate. At 4% that's 25 times, at 3.5% it's about 28.6 times, and at 3% it's about 33.3 times. It works in any currency, because it's just a ratio.

Only count the spending your savings have to pay for. Most countries pay some kind of state pension. If you'll get one, take it off first.

Say you spend $40,000 a year and a state pension pays $15,000. Your savings only need to cover $25,000. At 4% that's $625,000, not $1,000,000.

We tested every US retirement since 1928

We used Aswath Damodaran's yearly US returns from NYU Stern, which run from 1928 to 2025. Stocks are the S&P 500 with dividends. Bonds are 10-year US Treasuries. Inflation is the US consumer price index.

Each retiree starts with 60% in stocks and 40% in bonds and rebalances back to that mix once a year. They take the year's money out at the start of each year, because that's when you need it. There are no fees. Someone who retired in 1996 is the last to reach 30 years by 2025, so that's 69 retirements.

For each one, we found the highest starting rate that would have lasted the full 30 years. That's the number in the chart below.

In a typical year, you could have taken a lot more than 4%. The middle retiree could have started at 6.48%. Someone who retired in 1982, just before a long bull market, could have taken 10.26%.

But nobody knows in advance which year they're retiring into. The rule has to survive the worst year, and the worst was 1966, at 3.73%.

The four retirements that ran out

At 4%, the money lasted 30 years for 65 of 69 retirees. The ones who retired in 1965, 1966, 1968 and 1969 ran out between year 26 and year 28.

They retired just before 16 years of high inflation and weak markets. Prices nearly tripled between 1966 and 1981, rising about 196% in total. Inflation hit 13.3% in 1979.

Because the rule raises your income with inflation, their withdrawals nearly tripled too. Meanwhile, after inflation, stocks lost 14.5% over those 16 years and bonds lost 40.2%.

This is called sequence-of-returns risk. The same average return can end very differently depending on when the bad years come. Bad years early in retirement do the most damage, because you're selling while prices are low and there's less left to recover.

After 16 years, the 1966 retiree's $1,000,000 was down to $236,000 in 1966 dollars. It ran out in year 26.

Compare the 1929 retiree. They retired just before the 1929 crash and lived through the Great Depression. They still made it to 30 years, finishing with $524,000 in 1929 dollars. Falling prices in the early 1930s cut what they needed to take out.

Someone who retired in 2000, at the top of the dot-com bubble, is 26 years in. They've had two bear markets in their first nine years. They still have $648,000 in 2000 dollars, so they're on track.

Most retirees ended up richer

The rule is built to survive the worst case, so in most years it's far too careful. Of the 65 retirees whose money lasted, 43 finished 30 years with more than they started with, after inflation.

The middle one finished with 1.64 times their starting money in real terms. The 1982 retiree finished with 5.4 times. That's $5.4 million left over from a $1,000,000 start, in 1982 dollars.

That's the trade. A fixed 4% protects you from 1966. In most other years, you'd have spent less than you could have.

4% sits right on the edge

Here's how thin the margin is. We take each year's money out at the start of the year. Take it at the end instead, and the money lasts 30 years in 69 of 69 retirements.

That's one small change, and the four failures disappear. So in US history, 4% was about the most you could take, not a comfortable middle. Small changes in your fees, timing or mix can push you over.

Does your stock and bond mix matter?

Yes, a lot. Here's how many of the 69 retirements lasted 30 years at each rate, for five different mixes.

Too few stocks was the bigger danger. With 25% in stocks, 4% lasted only 51 of 69 times. Bonds didn't grow fast enough to keep up with rising withdrawals.

All stocks did about as well as 60/40 at 4% (64 of 69), with much bigger swings along the way. It also ran out once at 3.5%, which 60/40 never did.

At 4%, the best mixes were 75/25 and 60/40. At 5% and above, more stocks did better, but those rates ran out often whatever the mix.

What fees do to the 4% rule

Bengen's test had no fees. Real portfolios do. A cheap index fund might cost 0.1% or 0.2% a year. A financial adviser often charges about 1% a year on top of the funds.

A 0.2% fee changed nothing at 4%: still 65 of 69. A 1% fee dropped it to 58 of 69. With a 1% fee, even 3.5% failed four times.

So a 1% fee works out to roughly half a percentage point of withdrawal rate. If you pay one, plan for something nearer 3.5% than 4%.

Retiring early: 40 or 50 years

The 4% rule was built for 30 years. If you retire at 45 or 50, your money may need to last 40 years or more. The FIRE movement (financial independence, retire early) often uses 4% anyway. Here's how that held up.

Over 40 years, 4% lasted 50 of 59 times. Over 50 years it lasted 37 of 49. At 3%, the money lasted every time, over every length.

Outside the US, 4% was often too much

Everything above uses US data. The US had one of the best stock markets of the last century. That's partly why its history makes 4% look safe.

In 2010, the economist Wade Pfau ran Bengen's test on 17 countries, for retirements from 1900 to 1979. For each country, he found the highest rate that would have lasted 30 years from the worst starting year.

4% held in only four countries: Canada, Sweden, Denmark and the United States. In the UK the worst case was 3.77%. In six countries it was below 3%. A Japanese retiree in 1940 could only have taken 0.47% a year.

Pfau's numbers are generous, too. He let hindsight pick the best mix of stocks, bonds and cash for each country, and charged no fees. That's why his US figure is 4.02% while ours, with a fixed 60/40, is 3.73%.

A 2025 study by Anarkulova, Cederburg, O'Doherty and Sias went wider, to 38 developed countries. It found that a 65-year-old couple willing to accept a 5% chance of running out could take about 2.31% a year.

That wider data includes wars fought at home, bouts of hyperinflation and stock markets that shut for years. The US avoided the worst of those. If you invest mainly in your home country's market, its history matters more than America's.

So what rate should you plan around?

Nobody can tell you the right number for your situation, and we're not giving advice. But here's what the evidence above says, in one place:

• 4% has worked for a 30-year US retirement with low fees and 50-75% in stocks. It sat right on the edge, though.
• 3.5% lasted every 30-year US retirement, and 58 of 59 40-year ones.
• A 1% yearly fee costs about half a point. Take 4%, pay 1%, and the record falls to 58 of 69.
• Outside the US, the research points lower. The worst case was under 4% in 13 of the 17 countries Pfau studied and under 3% in six. The 38-country study put it at about 2.31%.
• In most years, a fixed 4% left a lot unspent. If you're willing to cut back after bad years, research on flexible spending rules finds you can start higher.

If you want the background, Stage 3 of the free course covers risk tolerance, time horizons and sequence-of-returns risk.

What this test can't tell you

• The 69 US retirements overlap. Someone who retired in 1966 and someone who retired in 1967 lived through mostly the same markets. There are really only about three separate 30-year stretches since 1928.
• Our bonds are 10-year Treasuries. Bengen used shorter-term government bonds, so his numbers differ a little from ours.
• We ignore taxes. They depend on your country and your account type, and they raise the amount you need.
• Withdrawals and rebalancing happen once a year. Monthly withdrawals would land between our start-of-year and end-of-year results.
• The future can be worse than the past. The countries in the international studies are the proof.