Common 401(h) Plan Mistakes to Avoid [Tips + Tricks]

A 401(h) account is one of the most efficient retiree-medical vehicles in the Code — deductible going in, tax-deferred while invested, tax-free coming out for qualifying medical expenses. But it is also unforgiving. Because the IRS has issued little formal guidance since the provision was enacted in 1962, and because these accounts are lightly administered relative to the pension plans that host them, mistakes are common — and they tend to fall into two costly buckets: qualification failures that put the entire host plan at risk, and funding traps that turn a tax advantage into a tax penalty.

Here are the errors that show up most often, and how to stay clear of them.

This is the single most expensive mistake, and it flows directly from how a 401(h) unwinds. Any money left in the account after all retiree-medical liabilities are satisfied must revert to the employer — and that reversion is punished twice. The returned amount is includible in the employer's gross income under §61 (subject to the §111 tax-benefit rule), and it is generally exposed to the §4980 reversion excise tax on top of that.

The base excise rate is 20%, but it climbs to 50% unless the employer establishes a qualified replacement plan or provides pro-rata benefit increases — mitigations that were designed for defined benefit pension surplus and don't map cleanly onto a leftover medical account. In practice, a private employer with a 401(h) residual is realistically looking at income tax plus a 50% excise on the same dollars.

The discipline that follows is simple: fund the 401(h) conservatively, toward expected actual retiree-medical spend, and top up rather than front-load. It is far better to slightly underfund and add later than to trap surplus in the account and watch half of it evaporate on the way out.

Section 401(h)(1) requires that medical benefits be subordinate to the retirement benefits, and the mechanical test — codified by OBRA '89 — is that aggregate contributions for medical benefits, added to any contributions for life insurance protection, cannot exceed 25% of total contributions to the plan.

Three features of this limit trip people up. First, it is an aggregate, cumulative test measured from the date the 401(h) account was established, not a fresh annual calculation — every year the running ratio of medical-to-total contributions since inception has to stay under the cap.

Second, it is reduced by any life insurance protection provided under the plan, so a plan carrying incidental life insurance has less room for medical funding than the headline 25% suggests. Third, and most commonly missed: in a year when the pension is fully funded and the employer makes no pension contribution, the subordination limit for that year is essentially 25% of zero.

You cannot feed a 401(h) that sits on a frozen or fully-funded plan with no current pension funding. If retiree-medical funding matters, the 401(h) needs to ride on a plan that is still actively funding the retirement benefit.

For any participant who is (or ever was) a key employee — which, in an owner-only plan, always includes the owner — the 401(h) benefits must be maintained in that individual's own separate "individual medical benefit account" under §401(h)(6). And once the account is individual, §415(l)(1) treats contributions to it as annual additions to a defined contribution plan for §415(c) purposes.

The consequence surprises owners: a solo owner's 401(h) contribution is capped at the defined contribution dollar limit ($72,000 for 2026, indexed), not at some larger "defined benefit" number — and it shares that ceiling with any other defined-contribution annual additions for the same person.

Designing an owner's 401(h) as if it were uncapped pension money, or ignoring how it stacks against a companion 401(k) or profit-sharing allocation, produces a §415 excess. The one piece of good news is that §415(l)(1) switches off the 100%-of-compensation prong for the individual medical account, so a modestly-paid owner isn't limited by low W-2 compensation — but the dollar limit still governs.

A 401(h) account may only pay the medical expenses of retired employees, their spouses, and their dependents. The definition of "retired" is where plans stumble. Under Treas. Reg. §1.401-14(b)(1), an employee is eligible if they are eligible to receive retirement benefits under the associated pension plan — and the IRS has confirmed (in PLR 202305001) that an active employee who is merely eligible for in-service pension distributions can qualify as "retired" for this purpose.

But the same ruling flags the trap: if separation from service is a condition of receiving pension benefits under the plan, then an employee still working is not retired, and the account cannot pay their medical expenses without jeopardizing qualification. Whether a given person can be paid turns entirely on the plan's own retirement-eligibility provisions, so those provisions have to be read carefully before any reimbursement is made.

Section 401(h)(2) requires that the medical benefits be funded through a separate account. This is an accounting requirement, not an investment requirement — the 401(h) assets can be invested alongside the general pension trust — but the plan must account separately for the medical contributions, earnings, and disbursements, and benefits must be payable only from the amounts allocable to that account.

Treating the medical money as an undifferentiated part of the pension fund, or failing to designate at the time of contribution the portion allocable to medical benefits, undermines the separate-account requirement and is a qualification defect for the whole plan.

Section 401(h)(3) requires that the medical benefits be reasonable and ascertainable, and the plan must specify the benefits and the terms on which they are paid. A plan that leaves the timing or amount of benefits to the employer's discretion fails this test.

Where the plan coordinates with other sources — a welfare benefit fund, a VEBA, or the employer's general assets — the document has to be specific about how the 401(h) benefits interact with those other sources. Vague or discretionary benefit language is a frequent and avoidable drafting failure.

Treas. Reg. §1.401-14(c)(2) prohibits a 401(h) arrangement from discriminating in favor of officers, shareholders, supervisory employees, or highly compensated employees — and that nondiscrimination test is applied across the retirement portion of the plan as well as the medical portion. The regulation names shareholders explicitly.

For a business with a genuine rank-and-file workforce, a proportionate 401(h) clears the bar easily. But an owner-heavy design that concentrates the medical benefit on the principals is exactly what the rule targets. The account can cover owners; it simply cannot be built to favor them.

The choice of host plan is usually made for front-end reasons — a cash balance plan generates large pension contributions, which under the subordination limit support a richly funded medical account, and the medical contributions are deducted separately under Treas. Reg. §1.404(a)-3(f), outside the §404(a)(7) combined limit. All true, and all attractive at inception.

What gets underweighted is the tail. A defined-benefit host requires an enrolled actuary and a Schedule SB every year for the life of the plan, even after the retirement benefit is frozen and fully funded — and you cannot shed that obligation without terminating the plan, which terminates the 401(h) and forces the reversion cascade described above.

Because an owner's retiree-medical account may need to pay out for decades, the host decision should be made with that multi-decade actuarial and administrative cost in view, not just the opening-year funding math.

Not every 401(h) error is one of over-reach; some are one of unwarranted caution. A common reflex is to assume that a more-than-2% S-corporation shareholder can't benefit from a 401(h) because §1372 treats such shareholders as partners for fringe-benefit purposes. But a 401(h) account is a feature of a §401(a) qualified plan, and a self-employed individual — including the deemed-partner shareholder — is treated as an employee for qualified-plan purposes under §401(c)(1).

Qualified retirement plan benefits are not among the fringe benefits §1372 disturbs. The >2% shareholder can participate in the 401(h) on the same footing as any other participant; the only real constraint is the nondiscrimination rule discussed above, which applies regardless of entity type. Writing off the 401(h) for an S-corp owner is a mistake built on a misapplied rule.

The pattern across all of these is that a 401(h) rewards conservative, well-documented design and punishes aggressive or casual design. The qualification mistakes — separate accounting, the "retired employee" definition, reasonable-and-ascertainable benefits, nondiscrimination — are matters of getting the plan document and its operation right.

The funding mistakes — overfunding, subordination, the §415(l) cap — are matters of respecting the limits rather than reaching past them. Get both right, and the account delivers exactly what it promises. Get them wrong, and the failure lands either on the plan's qualified status or on the employer's tax bill, neither of which is a comfortable place to discover the error.