The number that finally got my attention was $185,500. That's Fidelity's 2026 estimate of what a single 65-year-old will spend on medical costs over the rest of retirement — and read the definition before you budget from it. It is roughly 45% Part B and Part D premiums and about 48% cost-sharing. It excludes most dental care, and it excludes long-term care entirely, which are the two expenses most likely to break a retirement. It is a floor, not a total. Doubling it for a couple is the common shorthand, though Fidelity itself publishes only the individual number. I read that estimate at my desk one night while pulling together a Part D spreadsheet for my dad, and I had to put the laptop down for a minute.
I grew up watching my father, a process engineer for twenty-eight years, do retirement math the way he did everything: in pencil, in a composition notebook, every number checked twice. He did not estimate. He calculated. And when he and my mother actually retired, almost every assumption they had made in their fifties (about returns, healthcare, how long they'd live, when to claim) needed to be revisited. Some of the old rules of thumb held up. Some did not.
This is a piece about the budgeting choices that actually move the needle in retirement, with the current numbers I'd want my own parents working from. I am not a financial planner and I'm not the master of this yet. I am the son who sits at the kitchen counter with a calculator and asks: does this rule still work in 2026?
The 4% rule is having a public argument with itself
The 4% rule comes from a 1994 study by William Bengen. He tested every retirement start date from 1926 to 1976 against U.S. stock and bond returns running through 1992, and concluded that a retiree could withdraw 4% of their starting portfolio in year one, adjust that dollar amount for inflation each year after, and reasonably expect the money to last thirty years.
That rule got repeated so often that people forgot it was based on a specific window of history. Morningstar's retirement researchers, Christine Benz among them, have spent the last few years revising it. Their base-case starting rate for a 30-year horizon was 4.0% in 2023, 3.7% in 2024 and 3.9% in 2025, because today's bond yields and equity valuations make the next thirty years look different from Bengen's sample.
Meanwhile, Bengen himself has gone the other direction. He now puts it at 4.7% — but on a seven-asset-class portfolio, not the 50/50 his original paper used. The increase comes from diversification rather than from markets getting friendlier, which makes it less of a disagreement with Morningstar than it looks. Both things are in the air at once: the original author saying "actually, higher," and the institutional research saying "actually, lower."
Here's what I tell my parents, and what I'd tell you. The 4% number is a planning starting point, not a law. If you're starting retirement at 65 with a balanced portfolio, somewhere between 3.5% and 4.5% is a defensible withdrawal range to model. What matters more than the exact percentage is whether you're willing to flex the number down in a bad market year and back up in a good one. The retirees who run into trouble are the ones who keep withdrawing the same inflation-adjusted dollar amount through a 30% drawdown in their first three years. That's sequence-of-returns risk, and it's the thing that quietly breaks more retirements than anyone admits.
Social Security: the math on when to claim is closer than people think
The average retired worker now collects a little over $2,000 a month in Social Security, and the maximum benefit at full retirement age is $4,152 for 2026. Each year you delay past FRA up to 70, your benefit grows by roughly 8%. That's the delayed retirement credit. The 2026 COLA of 2.8% is already in those numbers — it took effect in January. You can pull your own projected numbers from the SSA Quick Calculator at ssa.gov.
The textbook advice has been: delay if you can. And the math, on paper, is real. Waiting from 62 to 70 increases your monthly check by about 77%, inflation-adjusted, for the rest of your life.
Granted, the break-even point matters. If you delay claiming, you're trading guaranteed checks in your sixties for larger checks later. The break-even point, where the bigger delayed benefit catches up to the smaller early one, lands around 80 and a half if you are comparing claiming at 62 against claiming at 70. Comparing full retirement age against 70 pushes it closer to 82. Those are different questions, and they get quoted interchangeably. If you have reason to believe you won't reach that age, delaying is not the obvious win.
But here's the part I think gets undersold. A 65-year-old couple in average health and not smoking has roughly a 50% chance that at least one spouse lives to 92. On general-population mortality the coin-flip age is closer to 90 — still far past where most people plan. That's not a small tail risk. It's a coin flip. And the higher earner's benefit becomes the survivor benefit when one spouse passes. Which gets me to the conversation I had with my own parents: if you can afford to delay the higher earner's claim, you're not just buying yourself a bigger check, you're buying your spouse a bigger floor for the years they may live alone. We have a separate piece on what happens to your spouse's Social Security when they die, and that's where the survivor math gets concrete.
If you're trying to think through your own claiming age, our walkthrough on choosing the right age for Social Security and Medicare breaks down the trade-offs in plain language.
Healthcare: the line item that ate the budget
I've already mentioned the Fidelity number, $185,500 per single retiree at 65 in its 2026 estimate. Here's how that decomposes for budgeting.
Standard Medicare Part B costs $202.90 a month in 2026, up from $185.00 in 2025. Verify at medicare.gov before plugging anything into a spreadsheet. If your modified adjusted gross income from two years prior exceeds the IRMAA threshold (above $109,000 single or $218,000 joint for 2026 premiums — those thresholds rose, so check the current year's before you assume you are over), you'll pay a surcharge on top, sometimes substantially. We have a current rundown of the 2026 IRMAA brackets and surcharge math.
Part D has finally gotten one piece of meaningful relief. Since 2025, thanks to the Inflation Reduction Act, there is a hard annual out-of-pocket cap on covered prescription drugs — $2,000 that first year, $2,100 in 2026, and $2,400 in 2027. For seniors on expensive medications (biologics, certain cancer drugs, insulin combinations) this is a real-money change after years of donut-hole exposure. The 2026 details are in our Part D prescription drug cap explainer.
Medicare Advantage premiums look cheap on the sticker (CMS puts the projected 2026 average at $14.00 a month, down from $16.40 in 2025) but the out-of-pocket exposure is higher than the premium suggests, and networks can be narrow. I am not anti-MA, but I am anti-treating the premium as the whole picture.
And then there's long-term care, which Medicare does not cover in the way most people assume. The data here is worth getting right, because it is quoted badly everywhere, including by me until recently. The famous "70% will need long-term care" figure comes from a 2005 study and counts any service, however brief. ASPE's current work puts it at about 56% who will have a significant need — help with two or more daily activities, or supervision for severe cognitive impairment, lasting at least 90 days. And most care is unpaid: across the whole population the average time spent using paid care is roughly 0.8 years, because 55% of people never use paid care at all. That average hides the thing that should actually worry you, which is the tail. A semi-private nursing home room runs a national median of $114,975 a year (CareScout, 2025 survey), so a few years of it takes everything most families have. Assisted living, where my own dad lives now, ran my family between $4,200 and $6,800 a month depending on level of care. If you have not at least looked at the basics of long-term care insurance, it belongs on your list this year.
The bucket strategy, in plain English
The 4% rule gives you a withdrawal rate. It does not tell you which account to pull from on Tuesday. That's where the bucket strategy earns its keep.
The simple version uses three buckets. Bucket one is one to two years of living expenses in cash and short-term Treasuries. Bucket two is roughly three to seven years of expenses in shorter-duration bonds. Bucket three is the long-term growth pile: diversified equities, longer-duration fixed income, whatever your risk tolerance supports.
The trick is that you spend from bucket one. When markets are flat or up, you refill it from buckets two and three on a normal cadence. When markets are down hard, the kind of year that breaks retirements built on a flat 4% withdrawal, you pause selling from bucket three and let bucket one and two carry you. That's how you defang sequence-of-returns risk without overcomplicating things.
Is it a perfect system? No. Critics correctly point out that the dollars are fungible: three buckets is really just a labeled asset allocation. Fair. But for actual humans making actual decisions in actual market downturns, having the buckets labeled changes behavior. My father, who could calculate flow rates in his head, still found the labels useful. That tells me something.
Tax-efficient withdrawal order, and the Roth window most people miss
The traditional withdrawal order is taxable accounts first, then tax-deferred (traditional IRA, 401(k)), then Roth last. The logic: let the tax-advantaged accounts compound as long as possible.
The more sophisticated approach, which has gotten a lot of attention in the last few years, is the "fill the brackets" strategy. In your early retirement years (after you stop working but before Social Security kicks in, and before Required Minimum Distributions start) your taxable income is often unusually low. That's the window. You can do partial Roth conversions, pulling money out of traditional IRAs and paying tax on it now while you're sitting in the 12% or 22% federal bracket, instead of watching it eventually fall into a higher bracket once RMDs and full Social Security stack on top.
Secure Act 2.0 pushed the RMD start age to 73, and to 75 for people born in 1960 or later. That's an extra one to three years of Roth conversion runway compared to the previous start age of 72. The penalty for missing an RMD also dropped. It's now 25% of the missed amount, reducible to 10% if you correct it within two years. Still steep enough that you don't want to miss one. We have a deeper piece on Roth conversions after retirement that walks through how to size them.
This is the kind of thing where, frankly, a few hours with a fee-only fiduciary advisor in your first year of retirement can pay for itself many times over. I say that as someone who is skeptical of the financial advice industry generally. Roth conversion math is one of the spots where the analysis is real, the stakes are high, and the answer depends on your specific brackets.
The real tools, and what to actually do this week
There are real tools. None of them are magic.
The SSA Quick Calculator at ssa.gov gives you a same-day estimate of your benefit at different claiming ages. Use it before you talk to anyone. The Medicare Plan Finder at medicare.gov is clunky but it is the source of truth for Part D plan comparisons in your ZIP code. Boldin (what used to be NewRetirement) and Fidelity's Retirement Planner are both serious modeling tools that handle Social Security, taxes, and withdrawal sequencing in one view. AARP's retirement calculator is friendlier but less precise. None of them substitute for a year of running your actual budget against your actual income.
If I had to name three things to do this week, they would be these. One: pull your most recent Social Security statement from ssa.gov and write down your projected benefits at 62, FRA, and 70. Two: estimate your retirement healthcare line (Part B, Part D, supplemental, plus a buffer) and put it on the budget as a fixed expense, because that's what it is. Three: if you're between 60 and the year your RMDs start, talk to a fee-only advisor about whether the Roth conversion window applies to you. That's a one-hour conversation that can save five figures.
My father still keeps a logbook. It used to be process adjustments at the plant. Now it's medication times, doctor appointments, and the running tally of what his assisted living has cost since he moved in. The instinct to write the numbers down does not retire when the job does. If anything, retirement is where the discipline pays off the most. I am still figuring out a lot of this myself, but I have learned that the people who do well are the ones who keep checking their assumptions against the latest data, and who are willing to update the rule when the rule deserves updating.
