How Is an RMD Calculated?

A required minimum distribution is one division problem. You take an account balance from the end of last year, and you divide it by a number from a table. The answer is the least you have to withdraw this year.

That is genuinely the whole mechanism. Treas. Reg. §1.401(a)(9)-5(a)(1) says the required amount “is equal to the quotient obtained by dividing the account balance … by the applicable denominator.” One number on top, one number underneath.

I am writing this page without any dollar amounts on purpose. The dollars change every year and the method does not. Once you can see the shape of the calculation, you can run it on your own statement in about a minute, and you can tell straight away when a custodian’s number looks wrong.

The two numbers, defined

An RMD is two inputs and one division.

The top number is your account balance on the last day of the prior calendar year. The bottom number is your applicable denominator, which comes from a life-expectancy table published in the regulations and depends on how old you turn this year.

Divide the first by the second. That is your required minimum distribution for the year.

Nothing in that calculation looks at what the market did this year, what you spent, what you earned, or what tax bracket you are in. It is deliberately mechanical. The tax code is not asking you to make a judgment call. It is asking you to move a specific amount of money out of a tax-deferred account so it finally gets taxed.

You can always take more than the required amount. Taking more this year does not reduce next year’s requirement, though. Treas. Reg. §1.401(a)(9)-5(a)(6) is explicit that “no credit towards a required minimum distribution will be given in subsequent calendar years for the excess distribution.” You cannot pre-pay.

Which account balance does the math start from?

The balance as of the end of the prior year. Not today’s balance. Not the average over the year. Not the balance on your birthday.

For an IRA, Treas. Reg. §1.408-8(b)(2) says to use “the account balance of the IRA as of December 31 of the calendar year preceding the calendar year for which distributions are required to be made.” For an employer plan, the parallel rule in Treas. Reg. §1.401(a)(9)-5(b)(1) uses the balance as of the last valuation date in the preceding year.

This is the single most useful thing to understand about the whole system, because it explains the timing complaint I hear most often.

Markets fall in the spring, and your required withdrawal for the year does not fall with them. It was already fixed by a balance taken months earlier. That feels backward, and it is worth knowing about in advance rather than discovering it in a bad quarter.

It cuts the other way too. A strong fourth quarter raises next year’s required amount, even if the account gives some of that back in January. The measurement date is fixed, so the number is fixed.

One practical note. Your custodian usually calculates this for you and shows it on a January statement or in the online portal. That figure is worth checking rather than trusting, because a custodian only sees the accounts it holds. It has no idea what is sitting at the other three firms.

What is the applicable denominator?

It is the number you divide by, and it is roughly a life expectancy in years.

The regulations used to call it the “distribution period,” and plenty of articles still do. The current language is “applicable denominator,” which I actually prefer, because it says exactly what the number does. It sits underneath the balance in a fraction.

The denominator gets smaller every year you age. That is the part that surprises people. A smaller denominator means a larger fraction of the account has to come out, so the required percentage rises steadily through your seventies, eighties, and nineties even if the balance is flat.

So the required distribution is not a fixed percentage of your account. It is a percentage that climbs, applied to a balance that moves. Two things changing at once is why the number rarely looks like last year’s.

There is a sensible logic underneath it. The denominator is built from mortality assumptions, and the tables were rewritten a few years ago to reflect people living longer. That rewrite lowered required withdrawals across the board, because a longer expected life means the same balance is spread over more years. It is the rare tax change that quietly helped almost everyone it touched.

The tables themselves have not moved since. What did move, twice, is the age at which you start using them.

Which life-expectancy table applies to you?

For almost everyone, it is the Uniform Lifetime Table.

Treas. Reg. §1.401(a)(9)-5(c)(1) sets the general rule: the applicable denominator “is determined using the Uniform Lifetime Table in § 1.401(a)(9)-9(c) for the employee’s age as of the employee’s birthday in the relevant distribution calendar year.” Your age this year, one row, one number.

I have kept the table itself off this page on purpose, because the method does not change and the figures do. The Uniform Lifetime Table in full sits on its own page, rendered from this site’s sourced-figure module, alongside the deadlines that go with it.

There is exactly one common exception, and it is a good one to know about because it lowers the required amount.

If your spouse is your sole beneficiary and is more than ten years younger than you, you use a different table. Treas. Reg. §1.401(a)(9)-5(c)(2)(i) sends you to the Joint and Last Survivor Table in §1.401(a)(9)-9(d) instead, using both your ages. Because it reflects two lives rather than one, the denominator is larger, and a larger denominator means a smaller required withdrawal.

The word sole is doing real work there. Treas. Reg. §1.401(a)(9)-5(c)(2)(ii) requires the spouse to be the sole beneficiary “at all times during the distribution calendar year.” Naming your spouse alongside a child, or alongside a trust, takes you back to the ordinary table. This is a beneficiary-designation detail with a real dollar consequence, and beneficiary forms are exactly the paperwork that quietly goes stale.

There is also a sensible allowance for a year when things change. If you were married on the first day of the year and the marriage ends during it, through death or divorce, Treas. Reg. §1.401(a)(9)-5(c)(2)(iii) says you do not lose sole-beneficiary treatment for that year merely because you were not married throughout it.

A third table exists, the Single Life Table, but it is for beneficiaries taking distributions from an account they inherited. That is a different calculation with a different clock, and I have written up the rule for inherited accounts separately.

Worth saying plainly why three tables exist rather than one. Each answers a different question about how long the money is expected to last. The Uniform Lifetime Table is built on a hypothetical beneficiary exactly ten years younger than the owner, which is why it applies as the default whoever your actual beneficiary is and whatever their age. The Joint and Last Survivor Table reflects two lives, one of them meaningfully longer. The Single Life Table reflects the beneficiary alone, after the original owner has died.

You do not choose between them. Your facts choose for you, and the facts are your birth year, your beneficiary designation, and whether the account is yours or inherited.

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

What age do you enter the table at?

Your age on your birthday in the year the distribution is for. Not your age when you take the money.

If you turn seventy-eight this coming November and take your distribution in February, you use the row for seventy-eight. The regulation reads on “the employee’s age as of the employee’s birthday in the relevant distribution calendar year,” so the whole calendar year uses one row.

The harder question is which year is your first year, and here is where a lot of published writing is now wrong.

The starting age is not one number. It is set by your birth year. Treas. Reg. §1.401(a)(9)-2(b)(2)(iv)–(vi) gives the applicable age as 73 for someone born on or after January 1, 1951 and before January 1, 1959, and 75 for someone born on or after January 1, 1960.

Read those two boundaries carefully and you will notice a gap. Birth year 1959 is not covered by either paragraph, because paragraph (v) — the one that would resolve it — is printed as “[Reserved]” in the regulation. The statute that created the change was drafted with conflicting provisions, and the regulation has not yet closed it.

So if you were born in 1959, your applicable age is genuinely unsettled in the published regulation. That is not a gap I can close for you, and neither can an article that states a single flat age with confidence. If that is your birth year, it is worth watching for guidance and worth asking your tax preparer what position they are taking.

For everyone else the rule is simple, and it is worth repeating because so much older material says otherwise. A flat starting age is wrong for roughly half of the people reading this. Check your birth year, not a headline.

And if you are still working, a workplace plan may start later than your birth year suggests. The applicable age is only half the test there. 26 U.S.C. §401(a)(9)(C)(i) sets the start at the later of the year you reach the applicable age and “the calendar year in which the employee retires,” so the plan at the employer you are still with can wait for the retirement. The exception is narrow in three specific ways. It does not touch an IRA — §401(a)(9)(C)(ii) rules it out for the IRA provisions by name — so an IRA starts on the ordinary schedule whatever you are doing for a living. It does not reach the old plans left behind at former employers, discussed in the next section, because they are not the employer you are still working for. And it is removed for an employee who is a five-percent owner of the business. It is also permissive rather than automatic: the plan document has to provide for it, so the answer comes from the plan administrator rather than from the birth-year table.

Which accounts can be combined, and which cannot?

This is where the mechanism stops being a single division problem and becomes a household one.

IRAs may be combined. Employer plans may not.

Treas. Reg. §1.408-8(e)(1)(i) says “the required minimum distribution from one IRA is permitted to be distributed from another IRA,” with the amount “calculated separately for each IRA” and the sum “distributed from any one or more of the IRAs.”

Read that twice, because it is a two-step rule and people usually remember only the second step. You still calculate each IRA’s required amount on its own. What you may then do is add those amounts up and satisfy the total from whichever IRA you like.

That flexibility is genuinely useful. It lets you take the whole household requirement out of the account holding the asset you wanted to trim anyway, and leave a concentrated position or an illiquid holding alone.

Employer plans do not work that way. A 401(k), a 403(b), or a 457(b) is its own plan, and the permission to satisfy one account’s requirement from a different account appears only in the IRA regulation. If you have two old 401(k)s from two former employers, each one owes its own distribution, taken from itself.

Two further boundaries that catch people out.

An IRA you inherited is not part of your own aggregation group. Treas. Reg. §1.408-8(e)(2)(i) limits aggregate treatment to “amounts in IRAs that an individual holds as the IRA owner,” and says an IRA acquired through someone’s death is not treated as one of yours. Inherited accounts stand alone.

And a distribution from an IRA cannot satisfy a required amount owed by an employer plan, or the reverse. They are separate systems that happen to share a formula.

This is the practical argument for consolidating old workplace accounts into an IRA before required distributions begin. Not because the tax result is different, but because four plans mean four separate withdrawals to track, each with its own deadline and its own way of going wrong.

One more boundary, and it is the one couples get wrong most. Your accounts and your spouse’s accounts never combine. Retirement accounts are individual, the calculation runs per person, and a joint tax return does not merge them. Each of you has your own balance, your own row in the table, and your own requirement. Taking extra from one spouse’s IRA does nothing for the other’s.

Do Roth accounts have a required distribution?

Not during your lifetime, if it is a Roth IRA.

Treas. Reg. §1.408-8(b)(1)(ii) is about as clear as regulations get: “No minimum distributions are required to be made from a Roth IRA while the owner is alive.”

So a Roth IRA never enters the calculation above. It is not part of your IRA aggregation group for this purpose, its balance is not on the top line, and it has no denominator.

The rule for a Roth account inside a workplace plan — a Roth 401(k) — used to be different, and older articles will tell you those accounts do carry a lifetime requirement. That changed, and the current regulation reflects it: Treas. Reg. §1.401(a)(9)-5(b)(3) excludes amounts held in a designated Roth account from the balance used in the calculation, for years up to and including the year of death.

After death is a different regime for both, and that is the inherited-account rules rather than this page.

The planning implication is the one worth sitting with. Every dollar you move from a traditional account to a Roth account is a dollar permanently removed from this calculation, for the rest of your life. That is a large part of why conversion work tends to happen in the window between retiring and required distributions starting, when income is often at its lowest.

What do people get wrong about this?

Four things get the RMD calculation wrong, and none of them are the arithmetic.

Using the wrong balance. Today’s balance instead of the one from December 31. The number moved, the requirement did not.

Using a flat starting age. Half the audience for this material has a different one, and the birth-year boundaries are unforgiving.

Assuming a workplace plan aggregates. Taking the household total from an IRA leaves the old 401(k) short, and being generous somewhere else does not fix it.

Trusting one custodian’s number as the whole answer. It is right about the accounts it holds and blind to everything else.

The calculation itself is a division problem you can do on a phone. What takes the care is knowing which balance, which table, which age, and which accounts belong in the same bucket.

If charitable giving is part of your year, there is one more piece worth reading, because a transfer sent straight from an IRA to a charity can count toward the required amount without ever entering your income. I have written up how a qualified charitable distribution works in full.