If you’ve done any reading about retirement, you’ve met the 4% rule: take four percent of your savings the first year, adjust it for inflation after that, and your money should last thirty years.
It’s clean, memorable, and fits on a napkin. It’s also the closest thing to a plan most people are ever handed.
Here’s what almost no one tells you: that rule was written in 1994. And whether it’s still “safe” is a genuinely good question — with an evidence-based answer.
Where it came from
In 1994, planner William Bengen ran the historical numbers and asked a narrow question: looking backward at U.S. markets, what’s the highest starting withdrawal rate that would have survived even the worst 30-year stretch? His answer was about four percent. Good work — but notice what it was. A question about a pile of money and a withdrawal percentage, built on a fixed 30-year window and a particular era’s market conditions. It said nothing about your taxes, nothing about what happens when a spouse dies, nothing about the order your returns arrive in.
The tell: even its author moved it
Here’s what should make you pause. The man who created the 4% rule doesn’t use 4% anymore. With more decades of data, Bengen himself now puts the safe starting figure closer to 4.7%.
Meanwhile, researchers at Morningstar — using forward-looking return assumptions instead of pure history — landed somewhere very different. Their 2026 figure is 3.9% (it was 3.7% the year before).
Sit with that. Two credible, respected sources. The same question. Answers almost a full percentage point apart — 3.9% on one end, 4.7% on the other. Not because anyone’s being dishonest, but because the answer depends entirely on the assumptions: history or forecast, this allocation or that one, this definition of “safe” or another. A number that swings that much depending on who’s holding the calculator was never a law of nature.
What “safe” quietly means
Look closer at that Morningstar number. It’s built on a 90% success rate. Read that again: the “safe” withdrawal rate carries, by its own definition, a one-in-ten chance of failure. Most people have never been told that the famous rules come with a built-in coin-flip’s worth of risk baked in. That’s not a scare tactic — it’s just the math, and you deserve to see it.
Why a rate was never a plan
Even if you found the perfect percentage, you’d still be missing most of the picture, because a withdrawal rate is silent on the three things that actually decide a retirement: the order of your returns (a bad first few years can break a plan a good average would have saved), your taxes (4% of a pre-tax IRA and 4% of a Roth are not the same money), and your spouse (nothing in the rule accounts for the survivor). A rule about a pile can’t do any of that. Only a plan for a life can.
So — is it safe?
The honest answer: “safe” depends on your numbers, your allocation, your taxes, and how much guaranteed income you already have. Somewhere between roughly 3.9% and 4.7% is a reasonable starting conversation for a balanced portfolio over 30 years — but it’s a starting point, not an answer, and it should flex with your real situation, not sit frozen from 1994.
If you want to know your number instead of a rule of thumb, that’s exactly what a reading of your actual plan is for.
Book Your Retirement Second Opinion
Bengen, “Determining Withdrawal Rates Using Historical Data” (1994) and his updated ~4.7% figure; Morningstar, “What’s a Safe Retirement Withdrawal Rate for 2026?” (3.9% at a 90% success rate, 30-year horizon). Full citations in Summit’s evidence base. Confirm current-year figures before publishing.
Educational only — not investment, tax, or legal advice, and not a recommendation of any specific product or strategy. Figures are sourced estimates, not projections or a promise of results. Securities offered through The Quantum Group, member FINRA/SIPC. Investment advisory services offered through Summit Global Investments, a Registered Investment Adviser. Insurance products offered through Summit Income Planning Group. Separate and unaffiliated entities.
