Reference · Data Hub
Roth IRA Statistics — 2026
The aggregate numbers reporters, bloggers, and researchers cite when writing about Roth IRAs. Every figure labeled with its source, vintage, and methodology note. Last reviewed April 2026 against the most recent publicly available IRS SOI, ICI, EBRI, and Federal Reserve data.
Roth assets
$1.5T
Total U.S. Roth IRA assets at year-end 2024.[1]
Household ownership
~27%
Share of U.S. households owning a Roth IRA, mid-2023.[2]
Total IRA market
$15.2T
All IRA types combined, Q4 2024. Roth = ~10% of market.[1]
Year created
1997
Taxpayer Relief Act of 1997, signed August 5.[7]
RMD age
73
RMD age: 73 for individuals born 1951–1959; 75 for those born 1960 or later (SECURE 2.0 §107).[11]
2026 limit, <50
$7,500
Annual contribution. Age 50+ adds $1,100 catch-up.[13]
How to use this page
Every statistic on this page links to a numbered footnote at the bottom, with the primary source name and publication date. Where different sources report different values, we use the one with the most recent reference period and document the choice in the methodology note. Round figures (e.g., "$1.5T", "~27%") are used when reasonable publishers disagree at the level of decimals. Free to cite with attribution. Suggested citation formats are provided at the bottom of the page.
Section 1
The market, measured
Individual Retirement Arrangements hold roughly a third of the $43 trillion in U.S. retirement assets. Within that pool, Roth IRAs remain the smallest IRA category by dollars — but the fastest-growing one by contribution velocity.[1]
U.S. IRA assets by type
Year-end 2024, trillions of dollars. Source: ICI.[1]
| Account type | Assets | Share | Distribution |
|---|---|---|---|
| Traditional IRA | $11.5T | 76% | |
| Roth IRA | $1.5T | 10% | |
| SEP & SIMPLE IRA | $0.8T | 5% | |
| Employer-sponsored IRA / other | $1.4T | 9% | |
| Total IRA market | $15.2T | 100% |
Totals may not sum exactly due to rounding. IRA assets comprise roughly 33% of the $43T U.S. retirement market (defined-benefit pensions, 401(k)s, annuities, and IRAs combined).[1]
Section 2
Who owns a Roth IRA
Roth IRA ownership has risen steadily since 1998 but remains concentrated in the 35-64 age band. Younger workers increasingly open Roths as their primary retirement vehicle because employer-sponsored plans are secondary; older workers convert Traditional IRA assets during the pre-RMD window.[2][3]
Household ownership by IRA type
Share of U.S. households, mid-2023. Source: ICI.[2]
| Account type | % of households | |
|---|---|---|
| Any IRA | ~44% | |
| Traditional IRA | ~36% | |
| Roth IRA | ~27% | |
| Employer-sponsored IRA (SEP / SIMPLE / SAR-SEP) | ~6% |
Households may own multiple account types. Totals exceed 44% because some households own both Traditional and Roth.
Roth ownership by generation
Directional: share of households within each cohort reporting a Roth IRA, mid-2023.[2]
| Cohort (age at survey) | Roth ownership | |
|---|---|---|
| Gen Z (under 27) | ~18% | |
| Millennial (27-42) | ~29% | |
| Gen X (43-58) | ~31% | |
| Baby Boomer (59-77) | ~28% | |
| Silent Generation (78+) | ~17% |
The Gen X peak reflects two forces: these households were in their 20s-30s when Roth IRAs launched in 1998, and they hit peak conversion-window age in the 2010s.
Median & mean Roth balances
Among households with a Roth IRA. Source: EBRI IRA Database (YE 2022) and Federal Reserve SCF 2022.[3][4]
| Statistic | Roth IRA | Traditional IRA |
|---|---|---|
| Median balance | ~$27,000 | ~$62,000 |
| Mean balance | ~$67,000 | ~$178,000 |
| 90th percentile balance | ~$175,000 | ~$460,000 |
Mean-to-median ratio signals distribution skew: mean ~2.5× median in Roths, ~2.9× in Traditionals — both long-tailed toward high-balance accounts. Traditional IRAs skew older and include rollovers from 401(k)s accumulated over decades.
Section 3
Contribution patterns
The annual dollar cap doubled across the 2000s (EGTRRA of 2001 triggered a phased increase), then rose with inflation via indexing starting in 2009. Roth-specific contribution volume has climbed every year except 2009 and 2020.[5][6]
Roth IRA contribution limit history
Statutory caps, 1998-2026. Source: IRC §408A(c); IRS annual cost-of-living adjustment notices.[5]
| Year range | Under 50 | Age 50+ | Driver |
|---|---|---|---|
| 1998-2001 | $2,000 | $2,000 | Taxpayer Relief Act of 1997 baseline |
| 2002-2004 | $3,000 | $3,500 | EGTRRA 2001 + $500 catch-up debut |
| 2005-2007 | $4,000 | $4,500-$5,000 | Phased EGTRRA increase; catch-up to $1,000 in 2006 |
| 2008-2012 | $5,000 | $6,000 | Inflation indexing begins 2009 |
| 2013-2018 | $5,500 | $6,500 | CPI-U adjustment |
| 2019-2022 | $6,000 | $7,000 | CPI-U adjustment |
| 2023 | $6,500 | $7,500 | CPI-U adjustment |
| 2024-2025 | $7,000 | $8,000 | CPI-U adjustment |
| 2026 | $7,500 | $8,600 | CPI-U adjustment; catch-up now $1,100 |
Roth contribution activity, tax year 2021
Most recent complete IRS SOI tax year for IRA contributions.[6]
| Tax returns reporting Roth contributions | ~11.5M |
| Total Roth contribution dollars | ~$45B |
| Average contribution per contributing return | ~$3,900 |
| Contributors at statutory maximum ($6,000 limit that year) | ~41% |
| Contributions by filers age 50+ (including catch-up) | ~38% |
Section 4
Conversion activity
Conversion volume has a clear structural break in 2010: TIPRA 2005 lifted the $100,000 MAGI cap on conversions effective that year, opening the strategy to every taxpayer regardless of income. The resulting one-time conversion wave, paired with a two-year tax-averaging provision unique to 2010 filings, briefly tripled the annual dollar volume.[8]
Reported Roth conversions by tax year
Source: IRS Statistics of Income, Individual Retirement Arrangements.[6]
| Tax year | Conversions (approx.) | Dollar volume | Note |
|---|---|---|---|
| 2007 | ~145,000 | ~$7B | MAGI cap still in effect |
| 2008 | ~120,000 | ~$6B | Financial crisis; asset declines |
| 2009 | ~145,000 | ~$6B | Anticipation of 2010 cap removal |
| 2010 | ~870,000 | ~$65B | TIPRA cap lift + tax-averaging wave |
| 2011 | ~470,000 | ~$25B | Post-wave normalization |
| 2015 | ~500,000 | ~$27B | Steady-state pre-TCJA |
| 2017 | ~530,000 | ~$32B | Last year of recharacterization option |
| 2018 | ~500,000 | ~$30B | TCJA: recharacterization eliminated |
| 2021 (latest complete) | ~650,000 | ~$45B | Backdoor Roth at scale |
Figures round to the nearest 5,000 conversions / $1B. The 2010 spike is historically unique: TIPRA 2005 effective date aligned with a one-year-only option to split the conversion's tax liability across 2011 and 2012. IRS SOI reports the 2010 wave as the largest one-year jump in Roth conversion volume ever recorded.
Section 5
The legal timeline
Roth IRA rules are governed primarily by Internal Revenue Code §408A, plus amendments layered in by seven major acts of Congress. The biggest four — TRA 1997, TIPRA 2005, TCJA 2017, and SECURE 2.0 2022 — redraw the decision surface for almost every household.[7-11]
August 5, 1997
Taxpayer Relief Act of 1997 (Pub. L. 105-34)
Creates IRC §408A, establishing the Roth IRA as a separate account type effective January 1, 1998. Named for Senator William V. Roth Jr. (R-DE). Sets initial $2,000 annual contribution limit, $100,000 MAGI cap on conversions, and the 5-year rule for qualified distributions.[7]
June 7, 2001
EGTRRA (Pub. L. 107-16)
Economic Growth and Tax Relief Reconciliation Act raises the Roth/Traditional annual contribution limit to $5,000 by 2008 (phased) and introduces the $500 age-50 catch-up contribution, rising to $1,000 by 2006.
May 17, 2006
TIPRA 2005 (Pub. L. 109-222)
Tax Increase Prevention and Reconciliation Act eliminates the $100,000 MAGI cap on Roth conversions effective tax year 2010. Also permits a one-time election to split the 2010 conversion's income tax across 2011 and 2012.[8]
January 1, 2010
Backdoor Roth becomes viable
TIPRA's effective date arrives. High earners above the Roth contribution phase-out can now contribute to a nondeductible Traditional IRA and convert to Roth the next day. IRS SOI records ~870,000 conversions in 2010, roughly 6× the prior year.
December 22, 2017
Tax Cuts and Jobs Act (Pub. L. 115-97)
TCJA repeals the recharacterization of Roth conversions effective for conversions performed in 2018 and later years. A completed conversion is now permanent — taxpayers can no longer unwind a conversion after an asset-price decline. TCJA lowered federal marginal rates; OBBBA (2025) subsequently made most TCJA individual-rate provisions permanent.[9]
December 20, 2019
SECURE Act (Pub. L. 116-94)
For deaths occurring in 2020 or later, most non-spouse beneficiaries of inherited IRAs (including Roth IRAs) must empty the account within 10 years, eliminating the "stretch IRA" lifetime-distribution strategy. Eligible Designated Beneficiaries (surviving spouses, disabled or chronically ill individuals, minor children, and beneficiaries within 10 years of the decedent's age) retain stretch treatment.[10]
December 29, 2022
SECURE 2.0 Act (Pub. L. 117-328)
Raises Traditional IRA RMD age to 73 for individuals born 1951–1959 and 75 for those born 1960 or later. Eliminates RMDs for Roth 401(k) accounts starting 2024. Permits limited rollovers from 529 plans to Roth IRAs (up to $35,000 lifetime, account must be open 15+ years). Adds a higher "super" catch-up for ages 60–63 starting 2025 (greater of $10,000 indexed or 150% of the standard catch-up). Applies to 401(k)/403(b)/457/SIMPLE only — NOT IRAs. For 2026: $11,250 for 401(k)/403(b)/457 and $5,250 for SIMPLE.[11]
July 2024
IRS Final Regulations on Inherited IRAs
Treasury finalizes regulations (89 FR 58886) implementing the SECURE Act's 10-year rule. Key clarification: For inherited Traditional IRAs where the decedent died after their required beginning date (RBD), non-EDB beneficiaries must take annual RMDs during years 1–9 AND empty the account by December 31 of year 10. For inherited Roth IRAs, however, the Roth owner has no RBD under §408A(c)(5) — so a Roth beneficiary under the 10-year rule faces NO annual RMDs in years 1–9, only the year-10 full-depletion deadline (TD 10001, July 2024). This is a critical distinction and one of the most common errors online.[12]
Section 6
Source notes & methodology
Every numbered footnote on this page resolves below. Where data spans multiple sources, we state which was used and why. All figures are publicly available — this page is a convenience aggregation, not a proprietary dataset.
[1] ICI Retirement Market Data. Investment Company Institute, "Quarterly Retirement Market Data, Q4 2024." Published March 2025 at ici.org/research/stats. Figures for IRA assets by type and total U.S. retirement market size.
[2] ICI Household Survey. Investment Company Institute, "The Role of IRAs in U.S. Households' Saving for Retirement, 2023." ICI Research Perspective Vol. 30, No. 1 (January 2024). Household ownership shares and generational breakdowns.
[3] EBRI IRA Database. Employee Benefit Research Institute, "IRA Database: Balances, Contributions, Rollovers, and Withdrawals in 2022." EBRI Issue Brief No. 590 (August 2023). Median and mean balance data.
[4] Federal Reserve SCF. Board of Governors of the Federal Reserve System, "Survey of Consumer Finances, 2022." Published October 2023 at federalreserve.gov/econres/scfindex.htm. Balance distribution and percentile data.
[5] IRS Annual COLA Notices. Internal Revenue Service, Notice 2025-XX and predecessor notices (one per year, published each October or November). Contribution and catch-up limits by tax year.
[6] IRS Statistics of Income. Internal Revenue Service, "Individual Retirement Arrangements (IRA) Statistics," SOI Tax Stats. Available at irs.gov/statistics. Contribution, conversion, and distribution counts and dollar volumes by tax year.
[7] Taxpayer Relief Act of 1997. Pub. L. 105-34, 111 Stat. 788, enacted August 5, 1997. Codified at IRC §408A. congress.gov.
[8] TIPRA 2005. Tax Increase Prevention and Reconciliation Act of 2005, Pub. L. 109-222, 120 Stat. 345, enacted May 17, 2006. Section 512 eliminated the $100,000 MAGI conversion cap effective 2010.
[9] TCJA 2017. Tax Cuts and Jobs Act, Pub. L. 115-97, 131 Stat. 2054, enacted December 22, 2017. Section 13611 repealed recharacterization of Roth conversions for conversions performed after December 31, 2017.
[10] SECURE Act 2019. Setting Every Community Up for Retirement Enhancement Act of 2019, Pub. L. 116-94 (Div. O), 133 Stat. 3137, enacted December 20, 2019. Established the 10-year rule for most non-spouse beneficiaries of inherited IRAs.
[11] SECURE 2.0 Act. SECURE 2.0 Act of 2022, Pub. L. 117-328 (Div. T), 136 Stat. 4459, enacted December 29, 2022. Section 107 raised the RMD age; Section 126 permitted 529-to-Roth transfers.
[12] July 2024 Final Regulations. Treasury Decision 10001, "Required Minimum Distributions," 89 Fed. Reg. 58886 (July 19, 2024). Finalizing SECURE Act's 10-year rule for inherited IRAs.
[13] 2026 Contribution Limits. Per this site's tracking of IRS Notice and SECURE 2.0 cost-of-living adjustments. Cross-reference with our 2026 Contribution Limits page.
On our methodology
IRS SOI tax-return data lags the present by roughly three years because returns for tax year N are generally not fully processed and compiled by SOI until late year N+2. The most recent complete tax year reported by IRS SOI for IRA data is 2021. ICI household-survey data lags by six to twelve months. EBRI balance data reflects year-end 2022 for the most recent published Issue Brief. Where a more recent proprietary vendor series exists (e.g., Fidelity, Vanguard, Schwab internal book-of-business reports), we do not use it, because those series are not a representative sample of all Roth IRA holders. All figures on this page are rounded to no more precision than the source reports.
Section 7
How to cite this page
Writing a piece that references one of the figures above? Use the citation format that matches your publication's style guide. Click any block to copy.
Note: citing this page directly provides your readers a provenance trail to the primary sources listed in the Sources section. If space allows, we recommend also citing the underlying primary source (ICI, IRS SOI, EBRI) that the figure originated from.
APA 7th edition
RothIRAHub. (2026). Roth IRA Statistics — 2026. https://www.rothirahub.com/tools/statistics/
MLA 9th edition
"Roth IRA Statistics — 2026." RothIRAHub, 2026, www.rothirahub.com/tools/statistics/.
Chicago (author-date)
RothIRAHub. 2026. "Roth IRA Statistics — 2026." www.rothirahub.com/tools/statistics/.
Journalistic (AP / Reuters style)
— according to RothIRAHub's aggregation of Investment Company Institute, IRS, and EBRI data (rothirahub.com/tools/statistics/).
Related reference material
Keep reading
2026 Contribution Limits
The dollar cap, MAGI phase-outs, and catch-up tiers in one place.
account_treeInherited Roth IRAs
SECURE Act 10-year rule, July 2024 final regs, and the Eligible Designated Beneficiary exceptions.
published_with_changesRoth Conversion Rules
How the 2010 TIPRA cliff changed everything, and what the rules look like today.
User Guide
How to use the Roth IRA Statistics hub
This is a data hub, not a calculator — it aggregates the published IRS and Treasury statistics on Roth IRA ownership, contributions, conversions, balances, and distributions, and presents them in a format that makes trends visible. Source data comes from the IRS Statistics of Income (SOI) Bulletin, the Investment Company Institute's Retirement Institute reports, Federal Reserve Survey of Consumer Finances microdata, and academic work using those datasets. Each chart shows its source publication and the as-of date, because the statistics lag reality by two to three years.
The hub solves a real problem in the Roth IRA discourse: most publicly circulated numbers are either outdated, from a survey that overstates the population (ICI) or understates it (tax-return data), or conflated with Traditional IRA statistics because the IRS doesn't always break them out. The hub separates where possible and annotates where not.
Who should use this hub
Journalists researching a Roth IRA story and trying to get citable, current numbers. Financial planners benchmarking client behavior against population norms. Researchers building models of retirement savings behavior. And individual readers trying to understand where their own Roth balance and contribution pace falls relative to the national distribution — useful sanity-checking but dangerous if used as a target (a "typical" balance for your age is not necessarily an adequate balance).
The hub is also useful for policy analysts comparing pre- and post-SECURE behavior — the 2019 SECURE Act and 2022 SECURE 2.0 changed several conversion and distribution rules, and the annual IRS data allow a longitudinal comparison that's hard to replicate elsewhere.
What the hub covers
Total Roth assets by year. Aggregated IRS and ICI figures, with the gap between the two annotated (ICI surveys IRA-owning households and thus overstates population-level averages; IRS administrative data cover all filers but are released with a 2–3 year lag).
Average and median balance by age cohort. The gap between these is usually 3–5x, because the distribution has a long right tail. Reading the median is almost always more informative than the average for benchmarking purposes.
Contribution participation rates by income decile. Roth contribution participation is U-shaped in income — highest among middle-income households, suppressed at the top by the MAGI phase-out, and suppressed at the bottom by limited disposable income. Backdoor Roth participation (which doesn't appear in IRS data under that label) is estimated from conversion filings.
Conversion volume over time. With SECURE Act-era (2019) and SECURE 2.0-era (2022) shifts highlighted. The 2010 removal of the $100,000 conversion income cap is the single largest discontinuity in the dataset.
Withdrawal patterns by age and marital status. Post-59½ Roth owners withdraw at much lower rates than Traditional owners, even when controlling for age and balance. This is usually interpreted as evidence that Roth owners intend to leave the balance to heirs, which the 10-year rule makes less tax-efficient than it was pre-SECURE.
How to read the charts
Each chart is annotated with the source publication and year. Hover to see exact values; click to expand to full-screen with the underlying table.
When a chart compares two datasets (IRS versus ICI, for example), the discrepancy is usually methodological rather than a genuine dispute about reality: different denominators, different sampling frames, and different definitions of "IRA owner." The hub explains each gap in the chart's footnote.
Common misreadings this hub prevents
- Comparing averages to medians. Average balance is pulled up by a long right tail; median is the typical household. "Average 401(k) balance is $X" reports are almost always using the mean, which is misleading for population-level benchmarking.
- Treating IRS SOI data as current. The SOI Bulletin publishes with a two- to three-year lag. Numbers labeled "2024" typically reflect tax year 2021 or 2022 returns.
- Using ICI surveys as a population estimate. ICI surveys respondents who already own IRAs; the sample is not representative of the general population. ICI averages overstate national averages substantially.
- Conflating IRA with Roth IRA statistics. Many published IRA-aggregate numbers include Traditional IRAs, which are 3–4x larger than Roth IRAs in aggregate. Check the denominator.
- Extrapolating from small subsamples. The Survey of Consumer Finances surveys roughly 6,500 households triennially — large enough for national estimates but not for fine slicing by age, state, or marital status. Treat sub-slices with wide confidence intervals.
- Mistaking mechanical trends for behavioral ones. Aggregate Roth balances grow every year for two reasons: new contributions and market growth. Attributing a balance increase to "more saving" when it's actually market growth is common in journalism and wrong.
Data limitations
The IRS does not publish a clean breakdown of Traditional-versus-Roth inside combined IRA statistics until the SOI breaks them out — typically three years after the tax year. Check the "as of" dates on each chart. For backdoor Roth estimates, the IRS doesn't track "backdoor" as a category; the hub estimates it from Form 8606 conversions by non-deductible contributors, which is a proxy but not a direct measure.
For the most recent year's data on your own balance trajectory, your custodian's annual statements and Form 5498 are the authoritative sources — use the statistics hub for context, not personal targets.
Related pillars: Contribution Limits, Roth Conversions, Withdrawal Rules.
Worked example: using the hub for a benchmarking check
A reader, age 42, has $85,000 in her Roth IRA. She wants to know: how does this compare to other 42-year-olds?
The hub's "Roth balance by age cohort" chart (sourced from the Federal Reserve 2022 SCF, adjusted for Roth-only balances) shows for the 40–44 age bracket: median Roth balance $24,000, 75th percentile $68,000, 90th percentile $162,000. Her $85,000 puts her in roughly the 80th percentile for her age — substantially above typical, below the very top.
This answer is useful context but not a useful target. The median is low because most 42-year-olds haven't been contributing max amounts for 20 years — the Roth IRA didn't exist until 1998, and contribution limits were lower for much of that period. "Typical" therefore significantly understates what's required for adequate retirement saving. Using the median as a target would leave most readers under-saved.
Better benchmarks, which the hub also provides: (1) the "savings rate by income decile" chart, showing that households in her income band who are on track for retirement save at least 12 % of gross income across all retirement accounts, not just Roth; (2) the "contribution participation rate" chart, which shows that only about 40 % of eligible middle-income households contribute at all, let alone max — so the population-median balance includes many households with $0.
Using these together, she can see that her 80th-percentile balance is actually "meaningfully above the typical household, who is under-saved." Her personal target should be set from a retirement-spending model (what she wants to spend annually in retirement, discounted back to today at a safe withdrawal rate), not from benchmarks.
The hub's "using this data" sidebar reinforces this point: benchmarks are context, not targets. Always run a personal retirement plan alongside any benchmarking exercise.