Key takeaways
- For the 2026 tax year the IRS caps 401(k) salary deferrals at $24,500 and total IRA contributions at $7,500. Counting employer money, one 401(k) can absorb $72,000.
- 96% of Vanguard's plans provided an employer contribution in 2024, and the average promised match was worth 4.6% of pay. No IRA has an equivalent.
- The average Vanguard plan offered 27.6 investment options in 2024. The Labor Department's own arithmetic puts a 1 percent fee gap at 28% of a 35-year balance.
- In bankruptcy, ERISA plan assets sit outside the estate with no ceiling. Contributory IRA assets are exempt only up to $1,711,975 for cases filed from April 2025.
- Deducting a traditional IRA contribution phases out between $81,000 and $91,000 of income for a single filer covered at work in 2026. The deferral has no income test.
They aren't competing for the same dollar, and three things actually differ
You're trying to work out whether the next dollar of retirement saving belongs in a workplace 401(k) or an IRA. Here's the honest shape of the answer: for most people it resolves into a sequence rather than a choice, and the sequence is set by rules, not preferences.
The tax mechanics are close cousins. Money goes in before tax or after, compounds untaxed, and gets taxed on the way out according to which flavour you picked. That part is almost the same in both accounts. Three things aren't.
First, a 401(k) can receive employer money. An IRA cannot, in any circumstance, because an IRA is your account and your compensation is the only source. Second, your employer picks the 401(k) menu; you pick the IRA's. Third, the two accounts sit in different places in federal bankruptcy law. Everything below works those three, using 2026 figures throughout.
The 2026 limits set the size of the question
Every number in this section applies to the 2026 tax year. These limits are indexed to inflation and move most years, so a figure you remember from an earlier year is probably stale.
The IRS caps employee elective deferrals to a traditional or safe harbor 401(k) at $24,500 in 2026. Total contributions across all your traditional and Roth IRAs are capped at $7,500. That's the headline gap, and it understates the real one.
The 401(k) carries a second, much higher ceiling. For 2026 the IRS raised the limit on total annual additions — your deferrals, your employer's contributions, and forfeitures allocated to your account — from $70,000 to $72,000. That's 9.6 times the IRA limit. Your own salary deferral can't fill it. Employer money can.
Catch-ups repeat the pattern. A participant aged 50 or over can defer an extra $8,000 in 2026 if the plan permits it, and the IRS says such a participant "generally can contribute up to $32,500 each year, starting in 2026". The IRA catch-up for the same person is $1,100, taking the IRA to $8,600. For anyone aged 60 to 63 during the year, the plan catch-up rises again to $11,250, which lifts the annual-additions ceiling to $83,250 — against $80,000 for everyone else using a catch-up. The chart above sets the six ceilings side by side, and the two tallest bars are the ones your own paycheque can't reach on its own.
Two constraints run the other way. Only $360,000 of your pay counts for 2026 when contributions are worked out, so a very high salary stops buying more match at that point. And from 2026, a participant whose prior-year wages with the plan sponsor exceeded $150,000 has to make catch-up contributions on a Roth basis. That removes the up-front deduction on the catch-up portion for higher earners. The IRA catch-up has no such rule.
The employer match is the one difference with no IRA equivalent
This is the part that isn't a tax argument at all. It's a compensation argument.
Vanguard's How America Saves 2025, drawn from its own recordkeeping book, reports that 96% of its plans provided an employer contribution in 2024. Half of them offered a match only, 36% offered both a match and a non-matching contribution, and 10% offered a non-matching contribution alone. The average value of the promised match was 4.6% of pay; the median was 4.0%.
Look at what that does to the total. The average participant deferred 7.7% of pay in 2024. The average total contribution rate, participant and employer combined, was 12.0%. The employer supplied roughly 4.3 percentage points of a portfolio's annual funding — more than a third of it — for no additional saving by the participant.
There's no IRA analogue to that. An IRA contribution is limited to your own taxable compensation, and nobody adds to it.
Two qualifications matter. The match is conditional money until it vests. The IRS is explicit that your own deferrals are "always 100% vested", while employer contributions to a 401(k) can sit on a schedule that runs out to six years. IRA-based employer plans, by contrast, "require that all contributions to the plan are always 100% vested". And the match is only available if there's a plan. In March 2025 the Bureau of Labor Statistics found retirement benefits available to 72% of private industry workers — 59% at establishments with fewer than 100 workers, against 90% at establishments with 500 or more.
Your employer picks the 401(k) menu; you pick the IRA's
A 401(k) menu is a fiduciary product. The Department of Labor requires employers to "establish a prudent process for selecting investment options and service providers" and to "select prudent and adequately diversified investment options". Someone else has done the choosing, and they're legally accountable for it. That's a genuine service, and it's also a constraint.
The average Vanguard plan offered 27.6 investment options in 2024. Counting each target-date series as one offering, the average was 17.5 and the median plan sponsor offered 16. Only 8% of plans offered more than 25 distinct options, and 9% offered 10 or fewer. Participants used an average of 2.3 funds. So the practical constraint is narrower than it sounds for most people, and absolute for anyone who wants something the menu doesn't carry.
An IRA has no menu in that sense. Your custodian's tradeable universe is the menu. That flexibility is worth something specific, and it's cost. The Labor Department's own worked example, in a fee booklet last revised in September 2019, runs a $25,000 balance for 35 years at an average 7 percent return. At 0.5 percent in fees the illustration ends at $227,000; at 1.5 percent it ends at $163,000. In the booklet's words, "the 1 percent difference in fees and expenses would reduce your account balance at retirement by 28 percent". That's an assumption-driven illustration rather than a forecast, but the arithmetic of fee drag is not in dispute — we've set out the same compounding in the 30-year compounding cost of fund fees, and the costs that sit outside the headline ratio in spreads, turnover and cash drag.
The direction of that gap isn't fixed. A large employer's plan can access institutional share classes an individual can't buy, and a small employer's plan can be markedly more expensive than a retail index fund. The point is that in an IRA the cost is your responsibility, and in a 401(k) it's someone else's — which cuts both ways.
IRAs aren't unrestricted either. The IRS's guidance on IRA contributions is clear that money in an IRA "can't be used to buy a life insurance policy", and that if an IRA invests in collectibles "the amount invested is considered distributed to you in the year invested". The menu is wide. It isn't everything.
Deductibility phases out for the IRA and doesn't for the deferral
Here's where the accounts stop resembling each other on tax. A traditional IRA deduction is income-tested when you or your spouse is covered by a workplace plan. For 2026 the IRS phase-out ranges are:
- Single or head of household, covered at work: $81,000 to $91,000.
- Married filing jointly, where the contributing spouse is covered: $129,000 to $149,000.
- Married filing jointly, where you aren't covered but your spouse is: $242,000 to $252,000.
- Married filing separately and covered: $0 to $10,000, and that band isn't indexed.
Two things about this are widely misread. The phase-out applies to the deduction, not the contribution — the IRS notes you can contribute to a traditional or Roth IRA even while participating in a workplace plan, but "you may not be able to deduct all of your traditional IRA contributions". A non-deductible contribution still shelters growth, and it still uses your $7,500.
The second point is the contrast. The IRS's 401(k) limits page sets out exactly two limits on your deferral: the $24,500 elective cap and the $72,000 annual-additions cap. Neither is an income test. A worker earning well above every band above can still defer the full amount pre-tax, subject only to the plan's own terms and its nondiscrimination testing.
In bankruptcy, ERISA plans have no ceiling and IRAs have one
This difference is statutory, and it's the one most comparisons skip.
Assets in an ERISA-governed 401(k) generally never enter a bankruptcy estate at all. The mechanism is two provisions read together. ERISA requires that "benefits provided under the plan may not be assigned or alienated". The Bankruptcy Code says a restriction on transfer of a beneficial interest "that is enforceable under applicable nonbankruptcy law is enforceable in a case under this title". In Patterson v. Shumate, decided in 1992, the Supreme Court held that "an antialienation provision in a qualified pension plan constitutes a restriction on transfer enforceable under 'applicable nonbankruptcy law'" for that purpose. There is no dollar cap in any of it.
IRAs get a different route. Retirement funds in an account exempt from tax under the relevant Code sections can be claimed as exempt, but the Bankruptcy Code caps the exemption for assets in individual retirement accounts at $1,000,000 as enacted, "without regard to amounts attributable to rollover contributions" and earnings on them. That figure is adjusted every three years. The most recent revision moved it from $1,512,350 to $1,711,975 effective 1 April 2025, a 13.2004 percent increase, and it applies to cases filed on or after that date.
Read the carve-out carefully, because it does most of the work. The cap bites on contributory IRA money. Money that arrived in an IRA as a rollover from a workplace plan sits outside the cap. So the practical difference for most households isn't the ceiling at all — it's that a large 401(k) balance keeps its unlimited treatment whether it stays in the plan or rolls to an IRA, while decades of direct IRA contributions are what the $1,711,975 counts.
All of this is federal bankruptcy law. Protection from ordinary creditors outside bankruptcy is a separate question governed largely by state law, and it varies. That's a real limitation on this section and it isn't covered here.
When the answer is both, the arithmetic of the combined limits
The two accounts aren't mutually exclusive, and the IRS says so directly: you can contribute to a traditional or Roth IRA even if you participate in a workplace plan. For 2026 that means $32,000 of contribution room for someone under 50 who can fund both to the cap, and $41,100 for someone 50 or over using both catch-ups.
What determines a sensible order is not opinion, it's which dollars behave differently from each other. Only one kind does: a deferral that triggers a match, because it's the only contribution that causes additional money to be added. On the Vanguard averages above, that's 4.6% of pay a matched participant receives and an IRA contributor doesn't. Above the match threshold, the deferral and the IRA contribution are competing on tax treatment, menu and cost rather than on free money — and the IRA deduction bands decide whether the traditional IRA's tax treatment is even available at your income.
The one thing the limits can't tell you is your own plan's match formula and vesting schedule. Those live in your Summary Plan Description, and they're the numbers that determine how much of the 4.6% average applies to you. Keeping the two accounts visible as one allocation — which is what LedgerTouch is for — is a separate problem from choosing between them, and the more common one once both are open. When contributions stop being the main driver of the balance is a different question again, and we've done the crossover arithmetic on it.
The strongest case against ranking the workplace plan first
The match argument sounds decisive. Several objections push back on it, and they aren't weak.
The match isn't universal. On Vanguard's own figures, 4% of plans provided no employer contribution in 2024, and the BLS access rates show the gap is concentrated at small employers, where 59% of workers had retirement benefits available. For someone in that group, the largest structural advantage of the 401(k) simply doesn't exist.
The match can also be taken back. Employer contributions to a 401(k) can sit on a vesting schedule that runs to six years, and someone who expects to leave within two or three years may be comparing a match they'd forfeit against an IRA dollar that's theirs immediately.
Cost is the sharpest objection. A match of 4.6% of pay is a one-off boost to a contribution; a fee difference is charged every year on the whole balance. The Labor Department's illustration shows a 1 percent gap costing 28 percent of the balance over 35 years. In an expensive small plan, the menu can eat a meaningful share of what the match added — and unlike the match, that charge doesn't stop.
Finally, the Roth catch-up rule from 2026 removes the up-front deduction on catch-up contributions for anyone whose prior-year wages exceeded $150,000, which narrows one of the deferral's advantages exactly where high earners had assumed it was largest.
What this evidence can't tell you
Start with whose data this is. The plan figures come from Vanguard's own recordkeeping book, not a census of US plans. Vanguard's client base skews toward larger employers, and its own 2024 rows are marked as estimates. The direction of any bias runs toward better plans than the average American worker has access to, which flatters the 401(k) side of this comparison.
The fee example has clear limitations too. It's a hypothetical in a booklet last revised in September 2019, and the 7 percent return in it is an assumption, not a forecast. It illustrates the mechanism of fee drag; it doesn't measure the fee gap between any particular plan and any particular IRA. We couldn't source a like-for-like average expense ratio for plan menus versus retail accounts from a primary source, so that comparison isn't made here.
The BLS access rates describe availability in March 2025, not generosity. A plan counts as available whether the employer contributes 1% or 10%.
And the whole piece is a description of account mechanics. It says nothing about your marginal tax rate now versus in retirement, state income tax, Roth versus traditional treatment inside either wrapper, or what happens if you need the money before 59. Those change the answer, and none of them are settled by a contribution limit. Sizing what you hold across both accounts as a single portfolio is a further step again, closer to the arithmetic of a retirement number than to this comparison.
What would change the conclusion
If your plan has no employer contribution — 4% of Vanguard plans in 2024 — the single largest structural difference disappears. The comparison collapses to menu, cost and creditor treatment, and on two of those three the IRA does at least as well.
If your plan's menu is expensive. The Labor Department's 28 percent figure sets the size of the hurdle. A match worth 4.6% of pay is a large advantage against a plan that costs a fraction of a point more than an IRA, and a much smaller one against a plan that costs a full point more.
If your income moves through a phase-out band. The single-filer range runs $81,000 to $91,000 for 2026 and shifts most years. Crossing it changes whether the traditional IRA deduction exists, without changing anything about the 401(k) deferral.
If the bankruptcy cap stops being adjusted. The $1,711,975 figure is fixed until the next three-yearly revision. Inflation between now and then erodes it in real terms, while the ERISA plan's unlimited treatment doesn't erode at all.
The number worth watching isn't the contribution limit. It's your plan's match formula and its vesting schedule — the two figures that decide whether the 401(k)'s one unique advantage is worth what it costs you in menu and fees, and the two the IRS publishes nothing about, because they're your employer's to set.