The Trinity Study, first published by three Trinity University finance professors in 1998, asked a specific and modest question: if a retiree withdrew a fixed percentage of a portfolio each year, adjusted for inflation, what withdrawal rate would have survived every rolling 30-year period in U.S. market history — including the Great Depression and the 1970s stagflation stretch? The answer, for a portfolio holding a meaningful stock allocation, came out close to 4%. That's the whole claim: it worked, historically, for 30 years, in the United States — not across markets, not across time horizons beyond that window. This piece is about what happens once you take that number and run it against 2026's actual data.
Start with the part most people skip: Trinity used U.S. stock and bond total-return data only. It never tested Japanese, European, or emerging-market sequences, several of which had multi-decade stretches worse than anything in U.S. history. A rule built entirely on one country's 20th century is a rule built on one draw from a much larger set of possible economic outcomes. That the U.S. draw happened to be good doesn't make it the base rate.
The 30-year horizon is the second constraint, and it matters more for this readership than the geography does. Trinity was modeling a 65-year-old retiring with three decades left. Someone retiring at 38 is underwriting a horizon closer to fifty or sixty years. The math wasn't built for that case, and nothing about it recalibrates automatically just because the retiree is younger.
Then there's the part that's moved recently. Morningstar's own December 2025 research put the safe starting withdrawal rate for a 2026 retiree, on a 30-year horizon, at 3.9% — not 4%. This represents an increase from their 2025 figure of 3.7%, reflecting slight improvements in capital-market assumptions and bond yields. These lower rates aren't about markets performing badly. The S&P 500 is up 18.9% over the trailing year as of mid-2026. It's about starting valuations — the price you're paying today for a dollar of future earnings. The original Trinity backtest already contains sequences that opened at high valuations; it sampled dozens of rolling 30-year windows across a century, including years that started near market peaks, so that flavor of sequence risk is already baked into the 4% finding. What it couldn't test is a starting point outside the range its own data covered. Today's valuations, by measures like the CAPE ratio, sit above most of the historical windows Trinity sampled from — which is why Morningstar's 3.9% functions as a patch for the backtest's blind spot rather than a rebuttal of its result.
Run the actual dollar difference. On a $1.25 million portfolio, 4% draws $50,000 in year one; 3.9% draws $48,750 — a $1,250 gap that sounds trivial. But flip the question: to draw that same $50,000 at a 3.9% rate instead of 4%, the portfolio has to be $1,282,051, not $1,250,000 — about $32,051 more principal, a 2.56% larger number. Anchor instead to the 3.7% figure from Morningstar's 2025 report, and the portfolio required for that same $50,000 climbs to $1,351,351 — 8.11% more than the 4% math would have demanded. A tenth of a percentage point in withdrawal rate is a rounding error in a spreadsheet cell and a five-figure gap in the size of the number you have to reach before you're allowed to stop working.
There's a partial offset worth naming, because it cuts against the doom framing. The 10-year Treasury sits at 4.57%, up meaningfully over the past month, and the 30-year is at 5.09%. Fixed income yields this level haven't been common for most of the past fifteen years. A 50/50 or 60/40 portfolio's bond sleeve can now carry more of the withdrawal load through actual yield than it could when the same allocation was earning next to nothing. Elevated equity valuations are pushing the safe rate down; better bond yields are pushing back the other way. The 3.7%-to-3.9% range is roughly where those two forces currently net out.
None of this touches the structural issue. Valuations move and yields move; the 30-year design horizon doesn't. A rule stress-tested against three decades of history was never going to be the right instrument for a retirement that might run twice that long, and no amount of the withdrawal rate drifting from 4.0% to 3.9% to 3.7% closes that gap. It's a different problem, solved by a different lever — usually a lower starting rate held flexible, or a plan built with the explicit expectation of adjusting spend in bad years, rather than a single fixed number carried unchanged for five decades.
The practical move isn't to memorize whichever percentage is currently in vogue. It's to treat 3.9% as this year's input to a model that should already have flexibility built into it — a variable withdrawal approach, or spending guardrails that tighten in down markets and loosen in good ones — rather than a fixed contract signed once at retirement and never revisited. And when you build the model, use return data, not folklore: the number that gets repeated in comment sections isn't always the number that appears in the actual research it's attributed to.
The Trinity Study answered a question about the United States and thirty years. FIRE, by definition, is asking about a longer horizon and hoping the same answer holds. It might. But that's a hope you should be pricing, not one you should be backtesting once and filing away.