2027 Roth IRA and IRA deduction phase-outs: the formula, closed

Every year, the same two questions arrive in September without anyone actually deciding the answer: what will the Roth IRA income limit be next year, and what will the traditional IRA deduction phase out at. Both get treated as mysteries the IRS reveals in the fall. They are not. The short answer: both phase-out families are indexed to the identical twelve-month inflation window that already set the 2027 tax brackets — not the third-quarter window that decides next year’s contribution limit — and with the Bureau of Labor Statistics’ August chained CPI now published, three of the four dollar ranges are arithmetic rather than forecast.

The clock nobody checks

Coverage of these numbers almost always assumes they move on the same schedule as the 401(k) and IRA contribution caps, because all of it lands in the same autumn news cycle. It does not. The Roth IRA phase-out is indexed under 26 U.S.C. § 408A(c)(3)(D), and the traditional IRA deduction phase-out under § 219(g)(8). Both provisions use the same borrowed language, adjusting their dollar amounts “by the cost-of-living adjustment determined under section 1(f)(3) for the calendar year in which the taxable year begins, determined by substituting ‘calendar year 2005’ for ‘calendar year 2016.’”

Section 1(f)(3) is the tax-bracket formula. It is the twelve-month average of the Chained Consumer Price Index for All Urban Consumers ending in August, the same window our 2027 IRS inflation adjustments tracker uses for the brackets, the standard deduction, the FSA cap, and the gift exclusion. The August 2026 reading — the twelfth and final month of that window — has already been published, which means the phase-out ranges close on the same day the brackets do, not on the September reading that decides the 2027 401(k) and IRA contribution limits. Those contribution limits run on a separate provision, § 415(d), measuring a third-quarter average that will not finish until September 30. Two different IRS numbers, both due in a November notice, arrive from two different BLS releases a month apart. Conflating them is the single most common error in coverage of these figures.

The base years differ from the tax-bracket window, though the mechanics are shared. Section 1(f)(3)(B) directs that a base year before 2017 be reconstructed by taking the regular CPI-U for the applicable window and multiplying it by the ratio between the 2016 chained index and the 2016 regular index — 0.569845. For “calendar year 2005,” that means the CPI-U average for September 2004 through August 2005, 192.7667, multiplied by 0.569845, giving a base of 109.8472. Both the Roth and IRA provisions use that identical 2005 base; nothing about the phase-out math differs between them beyond the dollar amounts they start from.

What the statute actually sets

The dollar figures being adjusted are written directly into the two sections:

Both statutes round the adjusted dollar amount to the nearest multiple of $1,000. The width of each range — $15,000, $10,000, or $20,000 — is fixed in the statute and is not itself adjusted for inflation; only the floor of the range moves.

Checking the method: four for four against the confirmed 2026 figures

Before trusting a formula to project 2027, it needs to reproduce 2026, which the IRS already set in Notice 2025-67. The 2026 window average — the same interim BLS figure the tax-bracket tracker used to reproduce the 2026 brackets exactly — is 177.11. Divided by the 109.8472 base, that is a ratio of 1.612332.

Four inputs, four outputs, four exact matches against the confirmed 2026 ranges ($153,000–$168,000 Roth single; $242,000–$252,000 Roth joint; $81,000–$91,000 IRA single; $129,000–$149,000 IRA joint). The same statute, applied to the same window, produces the numbers the IRS already published. That is what makes the 2027 run below more than a guess.

The 2027 numbers, both ways

The twelve months from September 2025 through August 2026 are now complete, except for one: the Bureau of Labor Statistics never published an October 2025 index, because data collection stopped during that year’s government shutdown. Averaging the eleven months that exist gives 182.8565; filling the gap with an estimate for the missing month, as our inflation tracker explains in full, gives 182.6293 instead. That difference of roughly a tenth of a percent is the only thing left undecided in this projection, and it moves the ratio from 1.66465 (eleven months) to 1.66258 (October imputed).

Three of the five rows land on the same rounded figure regardless of which reading the IRS uses, because they sit far enough from a $1,000 rounding line that a tenth-of-a-percent shift in the average does not cross it. The joint Roth range, and the spouse-not-covered IRA range, are the exception: $249,697.50 with the eleven-month average sits just above the $249,500 rounding line, while $249,386.55 with October imputed sits just below it — a gap of roughly $311 deciding whether the top of the range reads $259,000 or $260,000. The IRA joint figure came closest to flipping without actually doing so: $133,006 under the imputed reading is only $6 above the $133,000 floor, the tightest margin in the table, but not tight enough to change the outcome.

A different clock than the contribution limit — and the Saver’s Match

It is easy to read these ranges as if they travel together with the $8,000 IRA contribution limit projected for 2027, but they do not share a calendar. That figure is indexed under a third-quarter average of the regular CPI-U, a window that runs through September and will not close until the month is over; our 2027 401(k) and IRA contribution limit projection tracks that separate threshold. A filer whose MAGI is inside the phase-out range above is working against the $8,000 ceiling projected there, reduced by the fraction below — the phase-out sets how much of that ceiling is available, not the ceiling itself.

2027 is also the year the Saver’s Credit is replaced by the Saver’s Match, a separate, lower-income incentive with its own eligibility thresholds that are not the same as the Roth or IRA deduction phase-outs discussed here. The two systems apply to different income bands and should not be confused.

What the range means for a specific filer

The reduction inside the range is linear, and the statute spells out the arithmetic directly: the contribution limit is reduced by the same proportion that the excess of MAGI over the floor bears to the width of the range.

Take a single filer with a 2027 MAGI of $165,000, against the projected range of $158,000 to $173,000. The statute’s reduction ratio is the excess of MAGI over the floor, as a fraction of the range width: $165,000 − $158,000 = $7,000, and $7,000 / $15,000 = 46.7%. That is the fraction of the $8,000 limit the phase-out removes, leaving 53.3% intact: (1 − 0.467) × $8,000 = 8,000/15,000 × $8,000 = $4,266.67. The IRS worksheet in Publication 590-A applies its own further rounding to reach the exact dollar figure a filer would enter on their return, a mechanic this projection has not independently verified in the statute text and does not state as fact here — but the underlying fraction, roughly 53% of the limit remaining, is fixed by the statute regardless of that final rounding step.

This is also the point where the backdoor Roth strategy becomes relevant for anyone above the top of the range entirely. A filer with MAGI over $173,000 (single) or $260,000 (joint) cannot contribute to a Roth IRA directly at all, which is why the backdoor Roth IRA — a nondeductible traditional IRA contribution converted to Roth — exists as a workaround, and why the pro-rata rule matters enormously to anyone who has other pre-tax IRA money sitting around. The conversion step itself does not count against next year’s phase-out either: as confirmed above, § 408A(c)(3)(B)(i) explicitly excludes amounts included in gross income under the rollover-and-conversion provision from the MAGI used for this test. For the general definition of modified AGI and how it differs across programs, see the MAGI guide; for the underlying difference between the two account types these ranges apply to, see Roth vs. traditional IRA and the broader Roth IRA explainer. Anyone who contributes directly to a Roth IRA and later discovers their MAGI landed above the ceiling has a fix, and it is time-sensitive: see how to correct an excess Roth contribution.

The calendar from here

Nothing about the eleven-versus-imputed question resolves until the IRS publishes its revenue procedure. There is no announcement date confirmed for 2027 as of this writing; the closest precedent is Rev. Proc. 2025-32, issued October 9, 2025, for the 2026 figures, alongside Notice 2025-67 in November 2025 for that year’s 401(k) and IRA contribution limits. Watch the inflation adjustments tracker for the revenue procedure, which resolves the October question before the retirement-plan notice restates these ranges, and the 401(k) and IRA contribution limit projection for the separate September data point that settles the contribution ceiling itself.

Sources

• 26 U.S.C. § 408A(c)(3) — law.cornell.edu/uscode/text/26/408A
• 26 U.S.C. § 219(g) — law.cornell.edu/uscode/text/26/219
• 26 U.S.C. § 1(f)(3) — law.cornell.edu/uscode/text/26/1
• Bureau of Labor Statistics, Chained CPI-U and CPI-U series (SUUR0000SA0, CUUR0000SA0) — api.bls.gov/publicAPI/v2/timeseries/data/
• IRS Notice 2025-67 (2026 retirement plan cost-of-living adjustments) — cited for the confirmed 2026 control figures; document text not independently re-verified for this piece beyond the figures already reproduced in this site’s 401(k) and IRA contribution limit projection
• finbarrow, IRS 2027 inflation adjustments tracker — companion page for the shared CPI window and the October 2025 imputation question