The $10K Testwith JJ Free weekly email

Test #11 · Published November 1, 2026

Warren Buffett's 90/10 rule for retirement: we tested it on 21 retirees

By JJ, host of The $10K Test · · How the tests work

Watch on YouTube · Subscribe to the channel

My verdict: CONFIRMED, for retirees who take about 4% and can hold through a 50% drop without selling.

The short answer: Yes, at about 4% a year, in this test. Buffett's 90/10 mix never ran out of money for any of the 21 retirees who started from 1996 to 2016, taking 4% a year raised with inflation. But 100% stocks never ran out either, and ended with more money in 20 of the 21 start years: the 10% bond cushion paid off only for the one who retired in January 2000.

The key numbers

The claim

In his 2013 letter to Berkshire Hathaway shareholders, Warren Buffett described a bequest in his will: cash left to a trustee for his wife's benefit (his Berkshire shares go to philanthropic organizations). His instructions: put 10% of the cash in "short-term government bonds and 90% in a very low-cost S&P 500 index fund." He suggested Vanguard's, and said he believes the trust's long-term results will beat most investors who use high-fee managers. He gave no withdrawal rate.

Source: Warren Buffett, Berkshire Hathaway 2013 shareholder letter, dated February 28, 2014

How we tested it

Does the Warren Buffett 90/10 rule work for retirement?

At 4% a year, yes, in this test. All 21 retirees who started from 1996 to 2016 were still getting their checks in October 2026: with Buffett's 90/10, with 100% stocks and with 70/30.

Compare that with Dave Ramsey's 8%, which ran out for 12 of 13 retirees who started from 1996 to 2008, all in stocks (episode 1). Buffett's cushion does not rescue 8%: with 90/10, all 13 ran out.

Buffett gave no withdrawal rate, and his wife's trust is not a normal retirement account. The 4% and 5% are the test's choice.

Did the 10% in bonds help the retiree who started in 2000?

Yes, and that was the only one. January 2000 was the worst time to retire in this data: the dot-com crash, then 2008. With 90/10, the $1,000,000 fell to $346,964 by March 2009. Today it has $396,545, against $259,233 all in stocks. Same checks, same years: the cushion was worth +$137,312.

Per $10,000: $14,811 paid out so far either way, with $3,965 left in 90/10 and $2,592 all in stocks. Still going, not comfortable: $396,545 is about 5.1 years of checks at today's $77,259 a year. With T-bills the whole way instead of the bond fund, the cushion was worth +$78,771.

Why so little? 10% of $1,000,000 is $100,000: 2.5 years of spending at 4%. The S&P 500 with dividends was last below its January 2000 level in November 2011, about 11.9 years under water. What the cushion did do: bonds held or gained while stocks fell (the bond part gained 6.64% in 2008), and the January rebalance sold bonds to buy stocks after bad years.

Is 90/10 better than 100% stocks in retirement?

Usually not, in this test. In 20 of the 21 start years, 100% stocks ended with more money. The 1996 retiree, the only one with a full 30 years, has $9,772,655 all in stocks and $7,751,220 with 90/10: $2,021,435 less. Stocks grew 10.42% a year from January 1996 to October 2026; the bond part grew 2.49%.

The 2007 retiree shows both sides. At the March 2009 low, the 90/10 account was $47,344 higher ($518,310 against $470,966). Today it has $3,080,886, against $3,553,623 all in stocks: $472,737 less.

As for crashes: from October 2007 to March 2009, all stocks fell 55.2%, Buffett's 90/10 fell 50.2% and 70/30 fell 39.2%. With 90/10, a retiree still watched about half the money disappear.

Is 5% a safe withdrawal rate with a 90/10 portfolio?

Not for the unlucky start years in this test. At 5% ($50,000 the first year, raised with inflation), all stocks ran out for the 1999, 2000 and 2001 retirees. With 90/10, the 1999 and 2000 retirees still ran out, 13 and 14 months later (2000: January 2018 instead of November 2016). The 2001 retiree has $5,016 left, less than one month's check ($7,780). 70/30 ran out for all three.

The withdrawal rate mattered far more than the mix. For the 2000 retiree, the highest rate raised with inflation that never ran out through October 2026 was 4.1% all in stocks, 4.2% with 90/10 and 4.3% with 70/30.

One simple rule helped a little: after a year when stocks fell, take that year's checks from the bonds first and skip the rebalance. The 2000 retiree at 4% then had $514,017 left instead of $396,545. At 5% it bought only 4 more months.

$1,000,000 retired each January, 4% a year raised with inflation: what's left in October 2026
RetiredAll stocksBuffett 90/1070/30
1996$9,772,655$7,751,220$4,522,822
1997$7,162,793$5,792,821$3,512,841
1998$3,527,853$2,974,969$1,938,377
1999$1,318,570$1,233,389$932,814
2000$259,233$396,545$454,183
2001$1,864,517$1,666,708$1,204,104
2002$3,344,390$2,831,447$1,887,829
2003$6,591,771$5,310,518$3,257,191
2004$4,667,247$3,861,284$2,501,903
2005$4,276,122$3,601,715$2,427,541
2006$4,173,119$3,555,443$2,458,306
2007$3,553,623$3,080,886$2,204,674
2008$3,680,253$3,173,361$2,249,289
2009$7,958,353$6,312,295$3,843,214
2010$6,229,984$5,046,708$3,217,891
2011$5,484,069$4,497,580$2,945,200
2012$5,676,124$4,670,437$3,083,227
2013$4,852,821$4,060,923$2,780,667
2014$3,662,455$3,142,574$2,272,135
2015$3,206,888$2,794,219$2,087,553
2016$3,401,831$2,965,189$2,217,906

Money left is in October 2026 dollars. Only the 1996 retiree has a full 30 years; the 2009 to 2016 retirees began right after a crash and have 11 to 18 years of data.

The verdict

CONFIRMED for retirees who take about 4% and can hold through a 50% drop without selling. Buffett's 90/10 never ran out for any of the 21 retirees at 4%. But the 10% is insurance against retiring right before a crash, paid for in the good years: it paid off for 1 retiree in 21, it is not crash protection, and it does not make 5% safe.

What this test leaves out

Data: S&P 500 with dividends (SPY); bonds: 3-month T-bills to Jul 2002, then 1-3 yr Treasuries (SHY); CPI; no fees, no taxes. Every number on this page comes from the test results.

Questions people ask

What is Warren Buffett's 90/10 rule?
In his 2013 letter to Berkshire Hathaway shareholders, Buffett said the cash left for his wife's benefit should go 10% into short-term government bonds and 90% into a very low-cost S&P 500 index fund. He gave no withdrawal rate.
Does the 90/10 portfolio work for retirement?
In this test, at 4% a year raised with inflation, it never ran out for any of 21 retirees who started from 1996 to 2016. 100% stocks never ran out either, and ended with more money in 20 of the 21 start years.
Is 90/10 safer than 100% stocks in a crash?
A little. From October 2007 to March 2009, Buffett's 90/10 fell 50.2% and 100% stocks fell 55.2%. With 90/10, a retiree still watched about half the money disappear.
How much did the 10% in bonds help someone who retired in 2000?
It was worth +$137,312 on $1,000,000. Taking 4% raised with inflation, the 90/10 account had $396,545 left in October 2026, against $259,233 all in stocks.
Can you withdraw 5% a year with a 90/10 portfolio?
Not for the unlucky start years in this test. At 5% raised with inflation, the 1999 and 2000 retirees still ran out with 90/10, 13 and 14 months later than all stocks, and the 2001 retiree has $5,016 left.
What is the safe withdrawal rate for a 90/10 portfolio?
For the worst start in this test, January 2000, the highest rate raised with inflation that never ran out through October 2026 was 4.2% with 90/10, against 4.1% all in stocks and 4.3% with 70/30.

What's in the video

  1. 0:00 Intro
  2. 0:27 The claim
  3. 1:49 How we tested it
  4. 2:59 Four percent
  5. 4:12 The unlucky retirees, 2000 and 2007
  6. 5:54 Five percent, and the worst drop
  7. 6:55 Why it happened
  8. 7:46 What actually works
  9. 8:53 The verdict
  10. 9:26 Free weekly email + disclosure
Read the script of the video

The words JJ read on camera for this test. The final video may differ slightly.

Warren Buffett's plan for his own wife's money lost half its value in the financial crisis. And the 10% in safe bonds, the part that's supposed to protect you? For retirees taking 4% a year, it paid off for just 1 out of 21. Here's the twist: at 4%, his plan never ran out of money. Not once. Today we test Warren Buffett's 90-10 on 30 years of real data.

I am JJ, and this is the 10K Test. Every week I take a piece of money advice everybody repeats and test it against real market history. Not opinions. What the money actually did. Today's advice comes from Warren Buffett. In his letter to Berkshire Hathaway shareholders for 2013, he described instructions in his own will. His Berkshire shares are going to charity, so what he's leaving his wife is cash, run by a trustee. And he said his advice to that trustee couldn't be simpler. 10% of the cash goes in short-term government bonds. 90% goes in, in his words, a very low-cost S&P 500 index fund. He even suggested Vanguard's. And he wrote that he expects that to beat most investors who pay high-fee managers. People love this advice, and I get why. It's simple. It's cheap. And it comes from one of the most successful investors in history. The 10% in bonds sounds like a cushion for the bad years. But Buffett didn't say how much would be taken out each year, and his wife's situation isn't yours. So here's the real question: if a regular retiree uses Buffett's mix and lives off it, does it hold up? And does that 10% cushion really do anything? We tested it. Here's what happened.

Here's exactly what we did, so you can check it. 21 retirees. Each one retires with $1 million on the first trading day of January, one for every year from 1996 through 2016. Every month they take out their spending money. We tested two amounts. 4%: that's $40,000 the first year. And 5%: 50,000. Every year after that, the amount goes up with inflation, using the government's own inflation numbers. We ran three mixes. Buffett's 90-10. 100% stocks, like we tested in episode one. And 70-30, a bigger cushion. The stocks are the S&P 500 with dividends reinvested. For the short-term government bonds, we used 3-month Treasury bills until mid-2002, and after that a real fund of 1-to-3-year Treasury bonds. Every January, each account is put back to its mix. Real market history, month by month, right up to October 2026. Two things we left out, fees and taxes, and both would make every result worse. So this is the best case. And it's one 30-year stretch of one country's market.

Let's start with 4%, the number most retirees hear. Not one retiree ran out of money. Not with Buffett's mix, not with all stocks, not with 70-30. All 21 are still getting their checks in October 2026. Compare that with episode one. Dave Ramsey's 8%, all in stocks, ran out of money for 12 of the 13 people who retired from 1996 through 2008. And Buffett's cushion doesn't rescue 8%. With 90-10, all 13 ran out. Back to 4%. Here's the part that surprised me. In 20 of the 21 start years, all stocks ended with more money than Buffett's mix. The person who retired in 1996, the only one with a full 30 years, has about $9.8 million today with all stocks. With Buffett's 90-10, about 7.8 million. That's the price of the cushion in normal times. Stocks grew about 10.4% a year. The bonds grew about 2 and a half. $10,000 left alone in January 1996 became about 211,000 in all stocks, and about 177,000 in Buffett's mix. So why have a cushion at all? Because of the one year where it mattered.

January 2000. The worst time to retire in our data. The dot-com crash, then 2008. With Buffett's mix, this retiree's million fell to about $347,000 by March 2009. Today, all in stocks, they have about $259,000 left. With Buffett's 90-10, about 397,000. Same checks, same years. The 10% cushion was worth about $137,000. In $10,000 terms: every 10,000 paid out about 14,800 so far either way. All in stocks, about 2,600 is left. With Buffett's mix, almost 4,000. One honest note. If you use 3-month Treasury bills the whole way instead of that bond fund, the cushion was worth about 79,000, not 137. And even with Buffett's mix, 397,000, while taking out about 77,000 a year, is about 5 years of checks. Still going, not comfortable. Now 2007, right before the financial crisis. At 4%, nobody ran out with any mix. Here the cushion helped at the scariest moment. In March 2009, the all-stock account was down to about $471,000. Buffett's mix: about 518,000. About $47,000 less pain at the bottom. Then the market came roaring back, and the cushion held you back. Today the all-stock 2007 retiree has about 3,550,000. Buffett's mix, about 3,080,000. That cushion cost about $473,000.

Now 5%: $50,000 the first year. All in stocks, three retirees ran out of money: 1999, 2000 and 2001. With Buffett's mix, the 1999 and 2000 retirees still ran out, about a year later. The 2000 retiree made it to January 2018 instead of November 2016. 14 extra months. The 2001 retiree is technically still going, with $5,000 left. Their monthly check is about 7,800. So that's the end. Even 70-30, three times the cushion, ran out for all three. It bought the 2000 retiree a little over 3 years, to January 2020. And here's what the cushion did to the worst drop. From October 2007 to March 2009, the S&P 500 with dividends fell about 55%. Buffett's 90-10 fell about 50%. 70-30, about 39. So with Buffett's mix, you still sit there and watch half your money disappear.

Why did 10% do so little? Simple math. 10% of a million is $100,000. At 4%, that's 2 and a half years of spending. But the crash that started in 2000 didn't last 2 and a half years. The S&P 500, even with dividends, didn't get back above its January 2000 level for good until late 2011. Almost 12 years. A 2-and-a-half-year cushion can't carry you across a 12-year hole. What the cushion did do, it did quietly. From 2000 to 2002, and again in 2008, the bonds kept growing while stocks fell. And every January, putting the account back to 90-10 meant selling some bonds and buying stocks while they were cheap. That's where the 2000 retiree's extra money came from.

So what does work? Three things from the same data. One: the withdrawal rate matters far more than the mix. For the 2000 retiree, the most they could take, raised with inflation, without running out by today: 4.1% all in stocks. 4.2 with Buffett's mix. 4.3 with 70-30. The mix moved it by two tenths of a percent. Going from four to five is what sank people. Two: if you hold a cushion, spend it when stocks are down. We tested a simple rule: after a year when stocks fell, take your checks from the bonds first and don't rebalance. For the 2000 retiree at 4%, that left about 514,000 instead of 397. At 5%, it bought them only 4 more months. It helps. It doesn't rescue you. Three: don't sell in the crash. Buffett makes the same point in that letter: buy over time, and don't sell when the headlines are scary. Every number in this video assumes the retiree held on through a 50% drop. If you'd sell at the bottom, none of these results apply to you.

So, Warren Buffett's 90-10, used as a retirement plan. Verdict: CONFIRMED, for retirees who take about 4% and can sit through a 50% drop without selling. It never ran out for any of our 21 retirees. But don't mistake the 10% for safety. It cut the worst drop from about 55% to 50. At 4%, it paid off for 1 retiree in 21. And at 5%, it didn't stop the unlucky ones from running out. It's insurance against bad timing, and you pay for it in the good years.

Every week I send a free email with the full numbers behind each test, the tables, the charts, and what we're testing next. Free weekly email at 10Ktest.com. And tell me in the comments what money advice you want tested next. This video is for education and entertainment, not personal financial advice. Your situation is different; talk to a professional before changing how you invest or withdraw your retirement money. Disclosure: I don't own an S&P 500 index fund, a Treasury bond fund, or Berkshire Hathaway stock. I do own some individual stocks, including Apple, which is part of the S&P 500. Nobody paid for this video.

More tests

Every test →  ·  All videos →

Education and entertainment only. Not personal financial advice. Your situation is different; talk to a professional before changing how you invest or withdraw your retirement money. Disclosure: I don't own an S&P 500 index fund, a Treasury bond fund, or Berkshire Hathaway stock. I do own some individual stocks, including Apple, which is part of the S&P 500. Nobody paid for this video.