The 4% rule: what it means and what it leaves out

Key points

• The familiar 4% rule sets an initial annual portfolio withdrawal, then adjusts that dollar amount for inflation. It is not a fresh 4% of the remaining balance each year.
• A hypothetical $1,000,000 portfolio produces a $40,000 opening withdrawal. That is money leaving investments, not necessarily money available after tax.
• The original research studied particular U.S. stock and bond portfolios. Its historical results do not establish a future safe spending rate for every retirement.
• Withdrawal timing, investment costs, inflation, retirement length and the ability to reduce spending all change the question.

The useful part of the rule is its definition. It separates the first withdrawal from the later inflation adjustments. The dangerous shortcut is treating the resulting number as permission to stop checking the plan.

This guide explains that calculation and its original evidence. The retirement income guide places portfolio withdrawals alongside other income rather than treating savings as the entire household budget.

How does the 4% rule work?

In its familiar modern shorthand, the calculation has two steps:

1. Multiply the starting investment portfolio by 4% to set the initial annual withdrawal.
2. Adjust the previous annual dollar withdrawal for inflation, rather than multiplying the latest portfolio balance by 4% again.

For a hypothetical $1,000,000 portfolio, the starting amount is $40,000. If the inflation adjustment for the next withdrawal is 3%, the next amount is $41,200. A subsequent 2% adjustment produces $42,024.

The percentage of the remaining account can rise or fall. The dollar withdrawal follows the inflation rule while the investments follow the market. Those are different moving parts.

This is a total-return withdrawal method, not an instruction to earn that amount entirely from dividends or interest. Assets can be sold to fund the withdrawal. Nor is it a promise to leave the initial principal intact. Bengen’s research focused first on how long the portfolio lasted, with estate wealth a separate consideration. See the original paper’s portfolio-scenario discussion and conclusion, listed below.

The often-associated savings target of 25 times annual portfolio withdrawals is simply the reciprocal of 4%. It does not convert a budget into a finding that retirement is affordable. The definition of the budget still matters.

What did Bengen’s original study actually test?

William Bengen’s October 1994 paper examined U.S. common stocks and intermediate-term Treasury securities using Ibbotson historical data. Its initial illustration was a portfolio split equally between those assets. Other stock allocations were examined as well. The paper connected an initial 4% withdrawal, adjusted for inflation, with a minimum 30-year planning requirement in its tested framework.

A detail in the original appendix changes how the evidence needs to be described. The data ran through 1992, while the charts included retirement starts from 1926 through 1976 and displayed portfolio lives up to 50 years. Later portions were therefore not all observed history. Missing years were extended using assumed annual returns of 10.3% for stocks, 5.2% for bonds and 3.0% inflation.

The appendix also specifies withdrawals at each calendar year-end, annual rebalancing and taxes left out under a tax-deferred-account assumption. Its numerical illustration applies the first year’s inflation before that year-end withdrawal. It does not specify a separate advisory-fee deduction.

These details are visible in the publisher-hosted original scan, printed page 179. The publisher’s readable 2004 reprint omits that appendix. Neither the chart extensions nor the timing convention can be recovered by reading only a modern summary of the rule.

The distinction is not a reason to discard the study. It is a reason to describe what was actually measured, rather than calling every chart bar a completed historical retirement.

Did the Trinity study prove that 4% always works?

No. Philip Cooley, Carl Hubbard and Daniel Walz’s February 1998 AAII article used annual data from 1926 through 1995. It tested payout periods of 15, 20, 25 and 30 years, using the S&P 500 and long-term, high-grade corporate bonds. Those bonds were not Bengen’s intermediate-term Treasuries.

For the 30-year tests, the full sample contained 41 overlapping periods. In Table 3, which adjusts withdrawals for CPI inflation or deflation, an initial 4% withdrawal had a reported historical success rate of 95% for the equally split stock/bond portfolio and 98% for the portfolio with 75% stocks and 25% bonds. These are rounded results from that sample, not odds assigned to today’s reader.

The article also includes tables without inflation-adjusted spending. Reading one of those as if it maintained purchasing power changes the meaning of the result. Taxes and transaction costs were excluded, and success meant an ending portfolio value above zero, not a preserved estate or a comfortable margin.

The original article was checked through a disclosed PDF mirror because the AAII download was blocked. The source is the article itself, not the hosting site’s interpretation. The safe withdrawal rate guide explains why the definition of success is as important as the percentage.

What happens after an early market loss?

Consider a separate hypothetical example, independently calculated for this guide. It uses start-of-year withdrawals, unlike Bengen’s original appendix. There are no taxes, investment fees, contributions or outside income. Annual nominal portfolio returns are a loss of 20%, then gains of 10% and 5%. Inflation adjustments for the second and third withdrawals are 3% and 2%.

Plan year	Starting portfolio	Withdrawal	Ending portfolio
First	$1,000,000.00	$40,000.00	$768,000.00
Second	$768,000.00	$41,200.00	$799,480.00
Third	$799,480.00	$42,024.00	$795,328.80

The first row is: subtract $40,000, then apply the 20% loss to the remaining $960,000. In the second year, $41,200 is about 5.36% of the starting balance, even though the plan began at 4%.

If the second withdrawal were instead 4% of the new $768,000 balance, it would be $30,720. That is $10,480 less than the inflation-linked amount. It is a different spending rule, not a correction to the original calculation.

Timing also matters. Taking the same $40,000 only after the first year’s assumed 20% loss would leave $760,000, rather than $768,000. Neither convention is universally conservative across every return path. A comparison has to use the same convention on both sides.

These are invented returns chosen to expose the mechanics, not a market forecast or a reproduction of either original study. The sequence of returns risk guide isolates why changing the order of returns can change the remaining balance.

Does the withdrawal include taxes and fees?

An investment-account withdrawal and a spendable household income are not interchangeable.

If a household needs a hypothetical $40,000 after tax and every dollar of a particular withdrawal faces an assumed flat effective tax cost of 20%, the gross withdrawal would be $50,000. That arithmetic is not a tax estimate: actual taxation depends on account type, basis, other income and applicable law.

Investment fees are another drain. The SEC’s investor bulletin explains that costs leave less money invested to compound. Whether a model deducts them from returns, charges them against assets or includes them in the cash draw needs to be explicit. Counting a fee twice overstates the burden; leaving it out understates it.

Required distributions are a separate issue. The IRS describes RMDs as amounts that generally must leave covered retirement accounts. An RMD is not a finding that the household can sustainably consume that amount. The RMD guide covers that federal account requirement.

What would make the rule a poor match for a plan?

The comparison becomes incomplete when the household’s situation and the study’s assumptions differ. Useful questions include:

• Length: Is the modeled retirement shorter or longer than the period being discussed? Portfolio survival to a chosen final year says nothing about the following year.
• Assets: Are the holdings comparable to the research portfolio, or concentrated in a very different investment?
• Spending: Is the withdrawal inflation-linked, fixed in nominal dollars or tied to the current account value?
• Costs: Does the result include fund expenses, advice charges and the taxes associated with accessing the money?
• Flexibility: Can spending fall without leaving essential bills unpaid?
• Objective: Is finishing above zero enough, or is a reserve or bequest part of the requirement?

A flexible approach such as Guyton-Klinger guardrails changes the spending promise rather than making the original one more certain. Its cuts and freezes need to be evaluated as changes in living standards, not just improvements in portfolio survival.

What can a calculator establish?

The site’s historical withdrawal calculator compares a fixed inflation-linked spending plan with complete historical periods in its own stock, bond and inflation dataset. It includes a fee input and an outside-income input.

It does not reproduce Bengen’s original appendix or the Trinity tables. It is not a Monte Carlo model, and it does not implement Guyton-Klinger spending changes. It also does not accept the invented annual return path used in the table above through its public inputs.

Its results can illuminate the assumptions entered. They cannot turn an observed historical survival percentage into a future probability for a particular household. For an impending retirement after losses, retiring in a down market begins with the changed cash requirement and the portfolio available now.

About this guide

Written by the Diversify Guy editorial desk. Diversify Guy is a pen name. It is the byline of our editorial team, not a single individual. We do not publish personal trading records or claim to be your adviser. We tell you where every figure came from so you can check it yourself. How we work and how we get paid.

This is educational information, not individualized investment or tax advice. Research scope is stated with each U.S.-market study; examples are hypothetical. The RMD discussion concerns federal rules, not state tax. Evidence and calculator capabilities were checked November 3, 2026. Historical research is not a forecast for a retirement beginning on that date.

Sources and check dates

• William P. Bengen, October 1994, Determining Withdrawal Rates Using Historical Data: original paper and appendix, printed pages 171 to 180, especially 179. Publisher’s 2004 reprint checked for searchable body text, not the missing appendix. Checked November 3, 2026.
• Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz, February 1998, Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable: original AAII article, mirrored PDF, pages 16 to 21, especially Table 3. The direct publisher download was blocked. The mirror’s six pages match an archived AAII download in extracted text and rendered page pixels. Checked November 3, 2026.
• SEC Office of Investor Education and Advocacy, How Fees and Expenses Affect Your Investment Portfolio, also verified in an archived publisher copy when the direct download was blocked. Checked November 3, 2026.
• Jonathan T. Guyton and William J. Klinger, March 2006, Decision Rules and Maximum Initial Withdrawal Rates, withdrawal adjustments and purchasing-power trade-offs, PDF pages 5 to 12. Checked November 3, 2026.
• IRS, Retirement plan and IRA required minimum distributions FAQs. Checked November 3, 2026.