ACP Nondiscrimination Test: Calculation, Corrections, and Safe Harbor

The ACP nondiscrimination test, short for Actual Contribution Percentage test, compares employer matching contributions and employee after-tax contributions between higher-paid and lower-paid workers in a 401(k) plan. If matching and after-tax dollars skew too heavily toward the higher earners, the plan fails and the sponsor must correct the imbalance or risk losing its tax-qualified status. For the 2026 plan year, the line between the two groups sits at $160,000 in 2025 compensation.¹

What Contributions the Test Covers

Two categories of money flow into the ACP calculation. The first is employer matching contributions, including matches made under the SECURE 2.0 Act provision (effective for plan years beginning after 2023) that lets employers match an employee’s qualified student loan payments.²³ The second is voluntary employee after-tax contributions. These don’t reduce current taxable income the way pre-tax deferrals do, but the earnings grow tax-deferred inside the plan.

What stays out of the ACP is just as important. Pre-tax and Roth elective deferrals are tested separately under the companion Actual Deferral Percentage (ADP) test. Catch-up contributions for participants aged 50 and older (up to $8,000 in 2026, or $11,250 for those aged 60 through 63) are excluded from both tests entirely.⁴⁵ Every 401(k) plan that allows elective deferrals must pass both the ACP and the ADP independently. A plan can use different testing methods for the two, but doing so blocks certain correction techniques such as recharacterizing excess contributions or shifting qualified matching contributions between the two calculations.⁶ Most plans use the same method for both to avoid the complication.

Sorting HCEs From NHCEs

Before any math, every participant gets placed into one of two groups. The Highly Compensated Employee (HCE) group includes anyone who owned more than 5% of the business at any point during the current or preceding year, regardless of pay.⁷ It also includes employees whose prior-year compensation exceeded the IRS threshold, which is $160,000 (based on 2025 compensation) for the 2026 plan year.¹

The employer can optionally narrow the HCE group by electing the top-paid group rule, which limits HCE status to employees who also rank in the top 20% by compensation.⁷ Everyone else is a Non-Highly Compensated Employee (NHCE). Misclassifying even a handful of participants distorts every calculation that follows, so this step deserves careful checking.

Running the ACP Calculation

The test starts at the individual level. For each eligible employee, divide the total of matching and after-tax contributions by that person’s annual compensation. That produces an individual contribution percentage. Average all the individual percentages within the HCE group, then separately within the NHCE group.

The plan passes if the HCE group average stays within either of two statutory limits:⁸

125% test: the HCE average is no more than 125% of the NHCE average.
200%/2-point test: the HCE average is no more than two percentage points above the NHCE average, and no more than 200% of it.

Satisfying either one is enough. An example makes the second limit concrete. If the NHCE average is 2%, the HCE average can go as high as 4% (two points above, which also happens to be exactly 200%). If the NHCE average is 6%, two points above would be 8% while 200% would be 12%. The lesser controls, so the HCE cap is 8%.

One detail catches administrators off guard: eligible employees who chose not to contribute still count. Their individual percentage is zero, and that zero goes into the NHCE average. Skipping non-contributors inflates the NHCE average and makes the plan look more balanced than it is. Plans with weak participation among rank-and-file workers fail for exactly this reason.

Prior Year vs. Current Year Testing

Plans pick between two timing methods. Under the prior year method, the NHCE average from the previous plan year is compared against the HCE average from the current year. Under the current year method, both averages come from the same plan year. The prior year approach gives administrators a head start, because the NHCE number is locked in at the beginning of the year and the plan can project whether it will pass. New plans, and plans switching from current year to prior year testing, use the current year’s NHCE data for the first applicable year.

Correcting a Failed Test

When a plan fails, the sponsor has options, but the clock starts running on the last day of the plan year.

Deadlines

The first window closes 2.5 months after the end of the plan year. Excess aggregate contributions distributed to HCEs (with allocable earnings) before that date avoid the 10% excise tax that otherwise applies. For plans with an Eligible Automatic Contribution Arrangement covering all eligible employees, the window extends to six months.⁹

The absolute backstop is the end of the following plan year. If excess aggregate contributions haven’t been distributed or forfeited by then, the plan’s qualified status is in jeopardy.²¹⁰

Correction Methods

The most direct fix is distributing the excess and any allocable earnings to the HCEs whose contributions pushed the plan over. These refunds are taxable income to the HCE in the year distributed.

The alternative lifts the NHCE side of the equation instead. The sponsor contributes additional money to NHCE accounts until the NHCE average is high enough to bring the plan into compliance. Those contributions take two forms:

Qualified Matching Contributions (QMACs), which are extra matching contributions to NHCE accounts. They must be fully vested when allocated and carry the same withdrawal restrictions as elective deferrals.¹¹
Qualified Nonelective Contributions (QNECs), which aren’t tied to employee deferrals or matching formulas. Like QMACs, they must be fully vested on allocation and carry the same distribution restrictions.¹¹

Since 2018, sponsors can fund QMACs and QNECs from existing plan forfeitures rather than new cash if the plan document authorizes it, which cuts the out-of-pocket cost of a correction.¹²

Missed the Deadline? EPCRS Options

Blowing past the correction period doesn’t automatically disqualify the plan. The IRS Employee Plans Compliance Resolution System offers two later paths.⁴

The Self-Correction Program (SCP) lets the sponsor fix the failure without contacting the IRS. Within three years of the end of the correction period, SCP handles both significant and insignificant failures. After three years, SCP is available only for insignificant ones. The Voluntary Correction Program (VCP) requires a formal IRS submission, handles failures of any size at any time, and carries user fees of $2,000 to $4,000 depending on plan net assets.¹³

Under either program, the mechanics are the same: contribute QNECs to raise the NHCE average enough to pass, or distribute the excess to HCEs and deposit an equal dollar amount as a QNEC allocated to eligible NHCEs.⁴ The longer a failure goes uncorrected, the more expensive it gets, because earnings accrue on the excess amounts.

Skipping the Test With a Safe Harbor Plan

Sponsors who want to avoid annual testing can adopt a safe harbor 401(k). In exchange for guaranteed employer contributions to all eligible employees, the plan is deemed to satisfy both the ADP and ACP requirements automatically.¹⁴ The employer commits to one of three formulas:

Basic match: 100% of the first 3% of compensation deferred, plus 50% of the next 2%.
Enhanced match: at least as generous as the basic match, with the match rate not increasing as deferrals rise, and no match on deferrals above 6% of compensation.
Non-elective contribution: at least 3% of compensation for every eligible employee, whether or not they contribute themselves.

Traditional safe harbor contributions are 100% vested immediately. A Qualified Automatic Contribution Arrangement (QACA), a safe harbor design that includes automatic enrollment, may impose a two-year cliff vesting schedule on safe harbor matching contributions.¹⁴

Safe harbor status also comes with a notice obligation. Employers must deliver a written notice to each eligible employee at least 30 days (and no more than 90 days) before the start of each plan year, describing the matching formula, vesting schedule, and the employee’s right to change deferral elections.¹⁵ Employees who become eligible after the 90-day window opens must receive the notice no later than their eligibility date. Missing the notice deadline doesn’t automatically disqualify the plan, but it strips the safe harbor exemption for that year and forces the plan back into standard ADP and ACP testing.

1
Internal Revenue Service. Notice 2025-67: 2026 Amounts Relating to Retirement Plans and IRAs
2
Office of the Law Revision Counsel. 26 USC 401 – Qualified Pension, Profit-Sharing, and Stock Bonus Plans
3
Internal Revenue Service. Notice 2024-63: Guidance Under Section 110 of the SECURE 2.0 Act
4
Internal Revenue Service. 401(k) Plan Fix-It Guide – The Plan Failed the 401(k) ADP and ACP Nondiscrimination Tests
5
Internal Revenue Service. 401(k) Limit Increases to $24,500 for 2026
6
eCFR. 26 CFR 1.401(k)-2 – ADP Test
7
Internal Revenue Service. Identifying Highly Compensated Employees in an Initial or Short Plan Year
8
Office of the Law Revision Counsel. 26 USC 401 – Qualified Pension, Profit-Sharing, and Stock Bonus Plans
9
Office of the Law Revision Counsel. 26 USC 4979 – Tax on Certain Excess Contributions
10
Internal Revenue Service. Tax Consequences of Plan Disqualification
11
Federal Register. Definitions of Qualified Matching Contributions and Qualified Nonelective Contributions
12
Internal Revenue Service. Plan Forfeitures Used for Qualified Nonelective and Qualified Matching Contributions
13
Internal Revenue Service. Voluntary Correction Program (VCP) Fees
14
Internal Revenue Service. 401(k) Plan Fix-It Guide – 401(k) Plan Overview
15
eCFR. 26 CFR 1.401(k)-3 – Safe Harbor Requirements