The RMD Table for 2026, and the Deadlines That Go With It

Most people arrive at this page looking for one number. A row in a table, matched to an age, used once a year and then forgotten until next January.

That is a reasonable thing to want, and the table is below. But the number is the easy part. What actually costs people money is the calendar around it — which year is your first one, what date it is due, and what happens if a deadline goes by with nothing withdrawn.

So this page is the table plus the dates. The method behind the division — which balance, which table, which accounts combine — lives in a companion piece on how an RMD is calculated, and I am not going to repeat it here.

One note on the figures below before you use them. They are not typed into this article. They are read from the same sourced-figure module the rest of this site runs on, which is checked against a primary source on every build. A wrong divisor is a wrong required distribution, and I did not want that number to depend on me copying 49 rows correctly.

What is the RMD table by age?

It is called the Uniform Lifetime Table, it is published at Treas. Reg. §1.401(a)(9)-9(c), and it is the table almost every account owner uses.

Find the age you turn this calendar year in the left column. The number beside it is what you divide last year’s closing balance by. That quotient is the least you have to take out.

Source: Treas. Reg. §1.401(a)(9)-9(c). These figures are read directly from this site’s sourced-figure module, not retyped.

The third column is the same fact said the other way round. Dividing by a factor and taking a percentage of a balance are the same arithmetic, and most people find the percentage easier to hold in their head — it is the answer to “how much of this account do I have to spend this year”, which is the question underneath the divisor. It is computed here from the divisor beside it rather than typed, so the two columns cannot disagree.

Two things are worth noticing as you scan it.

The divisors shrink as the ages rise. That is how required distributions behave over a retirement: the same balance produces a larger and larger withdrawal, because the number underneath the line keeps getting smaller. A required distribution is not a flat percentage of your account. It is a percentage that climbs every single year, applied to a balance that moves on its own.

And the shrinkage is not even. The step down from one row to the next gets steeper as you move through the table. In your seventies the change from one year to the next is gentle enough that the required amount tracks whatever the market did. By your late eighties the divisor is falling fast enough to drive the number up on its own, even in a flat year. Households that plan around required distributions as a stable income line tend to be surprised by that second stretch, and it arrives at exactly the age when a spouse’s death or a move into care is changing everything else too.

What percentage of your IRA do you have to withdraw at each age?

Read the percentage column top to bottom and you can see the whole arc of it in one pass. The share required in your early seventies is small enough that a household spending broadly at that rate would barely notice the requirement existed. By the late eighties it has roughly doubled, and it keeps climbing from there — past one dollar in ten by the mid-nineties, and past one in seven at a hundred. That is the design working as intended: the tables are built to distribute the account across a life expectancy, not to preserve it. Any plan that treats the required amount as a ceiling on spending is quietly agreeing to a rising withdrawal rate for as long as it lasts.

How do you read the table on your own statement?

Two inputs, one division, and neither input is today’s.

Take your balance on the last day of last year. Find the age you will turn during this calendar year — not your age today, and not your age on the day you take the money. Read the divisor across from it. Divide.

I want to show you the shape of that arithmetic without putting a real divisor into a sentence, because a divisor in prose is a number nobody is checking. So here is the shape with a made-up round number, clearly labeled as such.

Suppose the balance was a million dollars, and suppose — purely as arithmetic, not as anyone’s actual row — the number underneath were a round 20.

A million divided by twenty is fifty thousand. That is the required minimum for the year.

If you would rather not do it on paper, the RMD calculator runs the same division against the same table, and it lets you enter more than one account so the aggregation problem below does not catch you out.

Now do it with your own two numbers. Your statement’s December closing balance on top, your own row from the table above underneath. It takes about thirty seconds, and the point of doing it yourself is not that your custodian gets it wrong. It is that your custodian can only see the accounts it holds, and it has no idea what is sitting at the other three firms.

That is the single most common way a household ends up short. Not bad arithmetic. Four correct numbers, from four institutions, none of which was looking at the whole picture.

What if your spouse is more than ten years younger than you?

Then the table above is the wrong one, and using it means taking out more than you have to.

There are two lifetime tables, not one. The Uniform Lifetime Table is the general rule, and Treas. Reg. §1.401(a)(9)-5(c)(1) states it as a general rule with an exception attached: it applies “except as provided in paragraph (c)(2) of this section (relating to a spouse beneficiary who is more than 10 years younger than the employee).” That exception, at Treas. Reg. §1.401(a)(9)-5(c)(2)(i), sends you to a second table — the Joint and Last Survivor Table at Treas. Reg. §1.401(a)(9)-9(d) — and it produces a larger divisor — which is to say a smaller required withdrawal — because it is measuring two lives instead of one.

Two conditions have to hold, and both are stricter than people expect.

Your spouse has to be more than ten years younger. Not ten. More than ten, measured by the ages you each reach during the distribution year rather than by your birthdays passing. A couple exactly ten years apart uses the ordinary table.

Your spouse has to be your sole beneficiary for the whole year. Treas. Reg. §1.401(a)(9)-5(c)(2)(ii) is unusually blunt about this: the spouse counts “only if the spouse is the sole beneficiary of the employee’s entire interest at all times during the distribution calendar year.” Naming your spouse for ninety percent and a child for the remaining ten does not partly qualify. It disqualifies the year outright. So does adding a contingent arrangement that makes somebody else a beneficiary of the interest during the year.

There is one piece of humanity written into it. Under §1.401(a)(9)-5(c)(2)(iii), a marriage that exists on January 1 but does not survive the year — a death or a divorce — does not retroactively cost you the treatment for that year. The change takes effect for the years after it, not the one it happened in.

Three practical notes follow from all of that.

This is not an election you make. If the conditions are met, the joint table is the applicable one; if they are not, it is not. There is no form and no choice, which means the mistake runs in both directions — households take too much because nobody told them, and occasionally take too little because a beneficiary designation quietly changed.

The divisor moves every year on two ages rather than one, so it has to be looked up annually. It is not a factor you find once and keep.

And the calculator on this site runs the Uniform Lifetime Table only. If this section describes you, read its answer as a ceiling — your actual requirement is lower — and get the joint figure from your custodian or your preparer rather than from a general-purpose tool.

If it helps to have the numbers in one place, I keep them all in the Ultimate Retirement Guide.

When is your first required distribution actually due?

Later than you would guess, and this is the part where a lot of published writing is now simply out of date.

The required beginning date is April 1 of the calendar year after the year you reach your applicable age. For an IRA, Treas. Reg. §1.408-8(b)(1)(i) puts it plainly: the required beginning date “is April 1 of the calendar year following the calendar year in which the individual attains the applicable age.” The parallel rule for employer plans is at Treas. Reg. §1.401(a)(9)-2(b)(1).

Those are not quite the same rule, and the difference matters if you are still working. The statute underneath them, 26 U.S.C. §401(a)(9)(C)(i), sets the required beginning date as April 1 of the calendar year following the later of the year you reach the applicable age and “the calendar year in which the employee retires.” That second limb is the still-working exception. The IRA rule quoted above has no equivalent to it, deliberately: §401(a)(9)(C)(ii) rules the exception out for the IRA provisions by name. Two further limits. The same clause removes it for an employee who is a five-percent owner of the business. And it only ever reaches the plan of the employer you are still working for, so an old plan left at a former employer starts on the ordinary schedule. And a plan can switch it off: Treas. Reg. §1.401(a)(9)-2(b)(4) lets a plan set one uniform required beginning date for every employee, which removes the exception for the whole plan. The default runs toward the exception; the plan is what takes it away. If you are past your applicable age and still at your firm, that is a question for the plan administrator rather than an assumption in either direction.

The words doing the work there are applicable age, because it is not one number.

The applicable age is 73 for people born in 1951 through 1958, and 75 for people born in 1960 or later. Treas. Reg. §1.401(a)(9)-2(b)(2)(iv) gives age 73 for an employee “born on or after January 1, 1951, but before January 1, 1959,” and paragraph (vi) gives age 75 for an employee “born on or after January 1, 1960.”

Read those two boundaries next to each other and you will find a hole between them. Birth year 1959 sits in neither. The paragraph that would resolve it, §1.401(a)(9)-2(b)(2)(v), is printed in the regulation as “[Reserved]” — the statute behind the change was drafted with provisions that conflicted, and the regulation has not closed the gap.

So if you were born in 1959, your applicable age is genuinely unsettled in the published regulation, and any article that hands you one confident number for that birth year is telling you something the source does not say. That is not a gap I can close for you. If it is your birth year, it is worth asking your tax preparer which position they are taking and watching for guidance.

For everybody else, the rule is easy once you stop looking for a single universal age. Check your birth year. A flat starting age is wrong for roughly half the people reading this, and it is wrong in the expensive direction for the younger half, who are being told to start two years early.

Why is the April 1 grace date a trap?

Because it moves the deadline, not the year.

Here is the mechanism, and it is worth reading twice. The regulation gives you until April 1 to take your first distribution. But that distribution still belongs to the earlier calendar year — the year you reached your applicable age. Treas. Reg. §1.401(a)(9)-5(a)(2)(ii) makes that first year a distribution calendar year in its own right, and §1.401(a)(9)-5(a)(3) says only that the distribution for it “may be made on or before April 1 of the following calendar year.”

Meanwhile, the second year’s distribution is running on its own schedule and is due at the end of that second year.

Use the grace period in full and both of them land in the same tax year. You take the first one in, say, March, and the second one in December. The regulation is satisfied. Your Form 1099-R shows two required distributions of taxable income arriving in one twelve-month filing period.

That is the trap, and it is not a small one for a household with a real balance. Two distributions stacked in a single year can push a couple through a bracket boundary they would otherwise have straddled comfortably over two. It can lift the income that determines your Medicare premium surcharge, which is measured on a return filed two years earlier and therefore reaches forward into a year you have not planned yet. It can pull more of a Social Security benefit into taxable income. None of those effects is a penalty. They are just the ordinary consequences of doubling one year’s ordinary income, and they are entirely avoidable.

The avoidance is simple: in most cases, take the first distribution in the year you reach your applicable age rather than deferring it into the following spring. One distribution per year, every year, no stacking.

There is one situation where deferring genuinely helps, and it is worth naming so this does not read as a blanket rule. Say you retire mid-year and your first partial-year income is unusually low while the following year is lower still. Or a large one-time item has already filled up the first year. In either case, pushing the distribution forward can be the better of two imperfect options. That is a calculation, not a default. Run it, or have someone run it, before the year ends — because after December 31 the choice has made itself.

When is every distribution after the first one due?

December 31, every year, with no grace period at all.

Treas. Reg. §1.401(a)(9)-5(a)(3) is explicit that the required minimum distribution for any distribution calendar year other than the first “must be made on or before the end of that distribution calendar year.” One sentence covers the rest of your life.

A few practical points that follow from it.

The deadline is the date the money leaves the account, not the date you asked for it. A request submitted on the twenty-ninth of December that settles on the second of January missed the deadline. Custodians know this and most publish an internal cutoff in early or mid-December for exactly that reason. Treat their cutoff as the real deadline and yours as the backstop.

The deadline is per account for employer plans and per person for IRAs. If you have old workplace plans from two former employers, each one owes its own distribution by the same date, taken from itself — that is the aggregation rule the companion article covers, and it is the boundary that catches the most people.

And nothing about the year-end deadline changes what the number is. The amount was fixed back in January by a balance struck on the last day of the prior year. Waiting until December to take it does not make it larger or smaller. It only means a year of market movement has happened between the measurement and the withdrawal, which is worth knowing if you are the kind of person who would rather not sell into a bad December.

What happens if you miss one?

An excise tax on the shortfall — and the correction rules are far more forgiving than most people believe.

26 U.S.C. §4974(a) measures the tax on “the amount by which such minimum required distribution exceeds the actual amount distributed during the taxable year.” Read that twice, because it is the part people get wrong in the frightening direction: the tax lands on the piece you missed, not on the account and not on the whole distribution.

The headline rate is not the rate most people end up paying. §4974(e)(1) cuts it for a taxpayer who corrects inside a window, §4974(e)(2) makes that window far longer than it sounds, and §4974(d) allows a waiver of the tax altogether for a shortfall “due to reasonable error” where “reasonable steps are being taken to remedy the shortfall.”

Those rules have their own page, because getting them right is procedural rather than arithmetical and the procedure is where the money is: what actually happens if you miss an RMD covers both rates, the exact shape of the correction window, Form 5329, and how the waiver is requested. One thing from it is worth carrying back here, because it is the sequence people reverse: take the missed distribution first, then file. Doing it the other way round is how a request for relief becomes an argument.

What should you actually do before year-end?

Four things, and none of them take long.

Confirm which year is your first one. Look up your birth year, not a headline age, and remember that 1959 has no answer in the regulation yet.

List every account that owes a distribution. Every traditional IRA, every old workplace plan, every account you have not logged into since the rollover you never finished. A required distribution you forgot about is the most common failure, and it is always an account rather than an arithmetic mistake.

Run the division yourself, once. Last year’s closing balance, your row from the table above, one division. Compare it against what each custodian is showing you. If they disagree, find out why before December rather than after.

Put the withdrawal in before your custodian’s cutoff, not before December 31. The two dates are not the same, and only one of them is enforceable.

If any part of your year involves charitable giving, there is one more move worth reading about first, because a transfer sent straight from an IRA to a charity can satisfy part of the required amount without ever passing through your taxable income. That is a different mechanism with its own age rule, and it is the rare piece of retirement tax planning that is strictly better than doing nothing.