Section 125 Nondiscrimination Testing: The Three Tests Explained for 2026

• Every cafeteria plan runs three independent tests under IRC Section 125(b): eligibility, contributions and benefits, and the 25% key employee concentration test.
• Unlike the separate IRC Section 105(h) self-insured medical plan test, Section 125 does not allow excluding part-time or seasonal employees from testing under IRC Section 125(g).
• A Dependent Care FSA inside the same cafeteria plan runs its own separate 55% average benefits test under IRC Section 129(d)(8).
• Employers with 100 or fewer employees can skip all three tests entirely through the IRC Section 125(j) simple cafeteria plan safe harbor with a minimum 2% nonelective or 6%-capped matching contribution.
• A failed test only strips pre-tax treatment from the highly compensated or key employees who caused the failure. Every other employee keeps full pre-tax treatment regardless of the outcome.

A 40-employee accounting firm in Columbus, Ohio built a Section 125 plan with a clean, identical election menu for every employee, ran payroll for three years, and never once tested it. When a new administrator finally ran the numbers, four partners holding a combined 62% ownership stake were electing $96,000 of the plan's $310,000 total pre-tax benefits, a 31% concentration ratio that had been failing the key employee test since year one. Nobody had lost money yet only because nobody had ever checked. Here is exactly how the three Section 125 nondiscrimination tests work for 2026, why a uniform plan document does not automatically mean a passing plan, and what actually happens when one of the three tests fails.

What is Section 125 nondiscrimination testing?

Section 125 nondiscrimination testing is the set of three annual compliance checks that IRC Section 125(b) requires every cafeteria plan to pass, designed to stop a plan from quietly becoming a tax shelter for owners and executives while providing little real benefit to the rest of the workforce. The three tests, the eligibility test, the contributions and benefits test, and the key employee concentration test, each measure a different kind of favoritism and each carries its own separate pass-or-fail outcome. A plan does not get to average its way to compliance. Passing two tests comfortably provides no credit toward the third, and a plan document that reads as completely uniform on paper can still fail based purely on what employees actually elected during the year.

Who counts as highly compensated for Section 125 testing purposes?

IRC Section 125(e) defines a highly compensated individual as an officer of the company, a person who owns more than 5% of the business, an employee who is highly compensated, or the spouse or dependent of anyone in those three categories. The statute does not fix its own dollar threshold for the "highly compensated" prong, so in practice nearly every third-party administrator applies the same $160,000 compensation figure used for 2026 under the separate IRC Section 414(q) highly compensated employee test, since building a second, competing compensation standard for the same population would create more confusion than it resolves. This population is distinct from the narrower key employee group used in the third test, and an employer needs both lists, not just one, to run all three tests correctly.

What is the Section 125 eligibility test?

The eligibility test under IRC Section 125(b)(1)(A) requires that a cafeteria plan not discriminate in favor of highly compensated individuals as to who is allowed to participate in the plan at all. A plan generally passes when its waiting period, minimum hours requirement, and any job classifications used to define eligibility apply the same way to every similarly situated employee, regardless of pay or title, and do not have the practical effect of covering the ownership and executive group while excluding most of the rank-and-file workforce. Written plan language that looks neutral can still fail this test if the actual eligible population, once the exclusions are applied, ends up skewed toward highly compensated individuals in practice.

Who a Section 125 plan can actually exclude from testing

IRC Section 125(g) permits excluding only four narrow categories of employees from the eligibility test: employees who have not attained age 21 before the close of the plan year, employees with less than one year of service, employees covered by a collective bargaining agreement where cafeteria plan benefits were the subject of good-faith bargaining between the union and the employer, and certain nonresident aliens with no U.S.-source income under IRC Section 410(b)(3)(C). Employers frequently assume the same part-time and seasonal exclusions available under the separate IRC Section 105(h) self-insured medical reimbursement plan test apply here too, but they do not. Section 125 has no part-time or seasonal carve-out at all, so a plan that quietly excludes part-time staff from cafeteria plan eligibility while covering every full-time officer is testing on a population that does not match what the statute actually permits.

What is the Section 125 contributions and benefits test?

The contributions and benefits test under IRC Section 125(b)(1)(B) requires that highly compensated participants not receive a disproportionate share of the plan's total nontaxable benefits or employer contributions compared to every other participant. A plan generally satisfies this test when every similarly situated employee is offered the identical menu of qualified benefits and the identical employer contribution formula, without a separate, richer election structure carved out for owners or officers. The test looks at both the availability of benefits, meaning whether the same options exist for everyone, and the actual utilization, meaning whether highly compensated participants are systematically receiving a larger average benefit even under a technically uniform plan design.

Worked example: passing on paper, testing on utilization

A 25-employee marketing agency offers the same $500-a-month employer contribution toward a Section 125 premium-only plan to every full-time employee, satisfying the availability half of the contributions and benefits test cleanly. But if the agency's three highly compensated officers also receive a separate, unwritten side arrangement letting them additionally elect an extra $300 a month in pre-tax dependent care benefits that no other employee is offered or told about, the plan now discriminates in actual contributions even though the written plan document never mentions a second tier. The fix is not more paperwork. It is removing the unequal arrangement so every similarly situated employee has access to the identical set of pre-tax elections, not just the identical base contribution.

What is the 25% key employee concentration test?

The key employee concentration test under IRC Section 125(b)(2) requires that the nontaxable benefits provided to key employees not exceed 25% of the total nontaxable benefits provided to every employee under the plan for the year. Key employee status is a narrower, separately defined population than highly compensated individual status, built on the officer, 5%-owner, and 1%-owner tests under IRC Section 416(i)(1)(A), and Benecor's key employee definition guide→ walks through exactly who qualifies for 2026. This test runs entirely independent of the two tests above it. A plan can offer a perfectly uniform election menu, pass the eligibility test cleanly, and pass the contributions and benefits test cleanly, and still fail this third test purely because a handful of key employees elected a disproportionate share of the actual dollars.

Worked example: the 25% test in dollars

A 40-employee accounting firm runs a Section 125 plan with $310,000 in total annual pre-tax elections across the entire workforce. Four partners who are all key employees under the ownership tests elect a combined $96,000 in pre-tax medical, dental, and dependent care benefits for the year. Dividing $96,000 by $310,000 produces a 30.9% concentration ratio, which fails the 25% limit by roughly $18,500 in excess key employee elections. To pass, the four partners would need to reduce their combined elections to $77,500 or below, the point at which $77,500 divided by $310,000 equals exactly 25%, or the excess amount becomes taxable wages to the four key employees who hold it, while the other 36 employees keep their full pre-tax treatment regardless of the outcome.

We had a completely identical plan document for every single employee and assumed that meant we were fine. Nobody told us the 25% test looks at what the partners actually elected, not what the plan document offered everyone else.

Does a Dependent Care FSA need its own separate test?

Yes. A Dependent Care Assistance Program, or DCAP, offered as a benefit inside a Section 125 cafeteria plan runs its own separate nondiscrimination test under IRC Section 129(d)(8), the 55% average benefits test, which is distinct from all three of the core Section 125 tests above it. This test compares the average dependent care benefit received by non-highly-compensated employees to the average benefit received by highly compensated employees, and requires the non-highly-compensated average to equal at least 55% of the highly compensated average. An employer that runs a clean eligibility test, a clean contributions and benefits test, and a clean 25% key employee test on the cafeteria plan as a whole can still have a DCAP quietly failing its own separate average benefits test underneath, since passing the parent plan's tests provides no protection for this fourth, independent check.

How does the simple cafeteria plan safe harbor avoid testing entirely?

IRC Section 125(j) lets an eligible small employer skip all three nondiscrimination tests, plus the separate underlying tests for group-term life insurance and dependent care benefits, by adopting a simple cafeteria plan instead of a standard one. Eligibility requires averaging 100 or fewer employees during either of the two preceding years, and an employer that grows past that count while already sponsoring a simple cafeteria plan can generally keep it in place until the workforce exceeds 200. In exchange for skipping the tests, the employer must commit to one of two minimum contribution formulas applied uniformly to every eligible employee who is not highly compensated or a key employee: either a flat nonelective contribution of at least 2% of each eligible employee's compensation regardless of what that employee elects, or a matching contribution of 200% of each employee's own salary reduction, capped once the employer's contribution reaches 6% of that employee's compensation.

The tradeoff behind choosing the safe harbor

The simple cafeteria plan safe harbor is not free. An employer that would otherwise pass all three tests on its own, without any minimum contribution requirement, gives up nothing by staying on a standard plan and simply testing every year. The safe harbor makes the most sense for a smaller employer with a concentrated ownership group, the exact profile most likely to fail the 25% key employee test, since the guaranteed minimum contribution to every eligible employee is traded against the ongoing risk and administrative cost of running three separate tests annually and correcting a failure after the fact. An employer already comfortably passing all three tests every year with room to spare is generally better off staying on a standard plan design rather than committing to a mandatory contribution formula it does not need.

What happens if a Section 125 plan fails nondiscrimination testing?

A failed test never disqualifies the entire plan or strips pre-tax treatment from employees who did not cause the failure. If the plan fails either the eligibility test or the contributions and benefits test, only the highly compensated individuals or highly compensated participants responsible for the discriminatory result lose their pre-tax treatment, and the affected amount becomes taxable income subject to federal income tax and FICA withholding, typically corrected through a W-2 adjustment. If the plan instead fails the 25% key employee concentration test, the consequence falls only on key employees, who must either reduce their elections to bring the ratio back to 25% or below, or include the excess amount as taxable wages. Every employee who is neither highly compensated nor a key employee keeps full pre-tax treatment on their own elections no matter which test the plan fails.

When should a Section 125 plan actually run these tests?

Most administrators run a projected test using current-year elections and compensation roughly two to three months before the plan year closes, since that timing leaves enough runway to correct a failing result, whether by adjusting a key employee's election or documenting a plan amendment, before the December 31 deadline that locks in the final numbers. A second, final test then runs after the plan year ends using actual full-year compensation and actual elections, because only that year-end data is legally determinative regardless of how the projected test looked months earlier. An employer that only tests once, at initial plan setup, is testing data that has nothing to do with whether the plan is compliant for the current year, since compensation, ownership, and actual elections all change annually.

Common Section 125 nondiscrimination testing mistakes

The most common mistake is assuming a uniform, identically written plan document automatically satisfies all three tests, when the key employee concentration test in particular looks at actual elected dollars rather than plan design. The second is confusing the highly compensated individual population used for the eligibility and contributions and benefits tests with the narrower key employee population used for the 25% test, and running only one list against both tests instead of building both correctly. The third is applying the Section 105(h) part-time and seasonal exclusions to Section 125 testing by mistake, understating the eligible population and producing a test result that would not survive an audit. The fourth is treating a passing cafeteria plan test as proof the Dependent Care FSA inside it also passed, when IRC Section 129(d)(8) requires its own separate 55% average benefits calculation.

How to run Section 125 nondiscrimination testing correctly

1. Pull last year's compensation and ownership data before open enrollment. Build both the highly compensated individual list and the key employee list separately, since they are not the same population.
2. Run the eligibility test on who can actually enroll. Confirm exclusions match only what IRC Section 125(g) permits, with no part-time or seasonal carve-out.
3. Run the contributions and benefits test on actual elections. Check utilization, not just the written plan document's availability language.
4. Run the 25% key employee concentration test separately. This test can fail even when the two tests above it pass cleanly.
5. Test the Dependent Care FSA on its own 55% average benefits test. A passing cafeteria plan does not mean a passing DCAP.
6. Correct before year-end or confirm the safe harbor instead. A projected test run two to three months early leaves time to fix a failing result before it becomes a taxable-wage problem.