401(k) Nondiscrimination Testing: ADP, ACP and Top-Heavy Tests Explained

9/26/2026

Every year, a traditional 401(k) plan has to prove that it does not favor its highest-paid employees. The proof is a set of annual nondiscrimination tests, and the most familiar of them, the ADP and ACP tests, the top-heavy test and the coverage test, are run by the plan's third-party administrator or recordkeeper. But the tests are only as good as the data they run on, and nearly all of that data comes from payroll.

That is why the people who know the most about why a plan failed its test are often in the payroll department, not the TPA's office. A failed test is usually a census problem, not a participant behavior problem. This guide explains how each test works, what payroll needs to deliver, why failures happen, and the deadlines that apply when they do.

Who Counts as Highly Compensated

All of the deferral and matching tests compare two groups: highly compensated employees (HCEs) and non-highly compensated employees (NHCEs). Getting this split right is the first step, and it is where many testing errors start.

Under Code section 414(q), an employee is an HCE for a plan year if they:

Were a 5% owner at any time during the current year or the preceding year, or
Had compensation from the employer above an indexed dollar threshold in the preceding year (the lookback year), and, if the employer elects, were also in the top-paid group, the top 20% of employees ranked by compensation.

The statute sets the base threshold at $80,000 and requires the IRS to adjust it for inflation. The IRS publishes the current figure each fall in its annual cost-of-living notice, so use the published number for the right lookback year instead of last year's.

Details That Trip Up the HCE Determination

Ownership attribution. For the 5% owner test, ownership is determined with the attribution rules of Code section 318. A spouse, child, parent or grandparent of a 5% owner is generally treated as owning that owner's shares, and is therefore an HCE regardless of pay. Payroll rarely knows these family relationships, so the plan sponsor has to supply them.
Lookback compensation. HCE status for the current year generally depends on the preceding year's compensation, so a new hire with a high salary is usually not an HCE in their first year under the compensation test.
Controlled groups. All employees of a controlled group or affiliated service group are treated as employed by a single employer. Leaving a related entity out of the census distorts both groups.
Former employees. Section 414(q)(6) can treat a former employee as an HCE if they were an HCE when they separated or after age 55.

The ADP Test

The actual deferral percentage (ADP) test compares the average deferral rates of the two groups. Each eligible employee's deferral rate is their elective deferrals divided by their compensation. Eligible employees who deferred nothing count as 0%, which is why low participation pulls the NHCE average down.

The test is passed if the HCE average meets either of these limits under Code section 401(k)(3)(A)(ii):

25 test: the HCE ADP is not more than the NHCE ADP multiplied by 1.25, or
2-point test: the HCE ADP exceeds the NHCE ADP by no more than 2 percentage points, and is not more than twice the NHCE ADP.

NHCE ADP	Maximum HCE ADP
1%	2%
2%	4%
3%	5%
4%	6%
6%	8%
8%	10%
10%	12.5%

The table shows how the practical cushion works. When the NHCEs average about 2% to 8%, the HCEs can generally average no more than 2 points above them.

Prior-Year vs. Current-Year Testing

The statute compares the HCE ADP for the current plan year with the NHCE ADP for the preceding plan year, unless the plan elects to use the current year for both groups. Prior-year testing lets the plan know the NHCE benchmark at the start of the year and manage HCE deferrals against it. Current-year testing uses actual current-year data for both groups and can produce a better or worse result depending on how NHCE participation changes. The method is set in the plan document, and switching it has restrictions, so confirm what your plan uses before you plan around it.

What Counts in the ADP Test

Pre-tax and Roth elective deferrals both count. Catch-up contributions by participants age 50 and over generally do not count in the ADP test. That is because catch-ups are, by definition, deferrals above a limit, including the ADP limit. Employer matching contributions do not count in the ADP test. They are tested separately in the ACP test.

The ACP Test

The actual contribution percentage (ACP) test applies the same two limits to employer matching contributions and employee after-tax contributions. Each eligible employee's contribution rate is their match plus after-tax contributions divided by their compensation. Employees who receive no match because they did not defer count as 0%.

The ACP test frequently fails in plans where HCEs defer at high rates and therefore collect the full match, while NHCEs defer little and receive little match. A plan with a generous match formula and low NHCE participation can fail the ACP test even when it barely passes the ADP test.

The Top-Heavy Test

Top-heavy testing works differently. It does not compare contribution rates. It looks at account balances.

Under Code section 416(g), a defined contribution plan is top-heavy for a plan year if, as of the determination date (generally the last day of the preceding plan year), the total account balances of key employees exceed 60% of the total account balances of all employees. Related plans of the same employer may be aggregated.

Key Employee vs. HCE

Key employees are a different group from HCEs, defined in section 416(i)(1). An employee is a key employee if, at any time during the plan year, they were:

An officer with compensation above an indexed threshold (the statutory base is $130,000; the IRS publishes the current figure each year). No more than 50 employees are treated as officers, and fewer in small companies.
A 5% owner, or
A 1% owner with compensation of more than $150,000.

The Consequence of Being Top-Heavy

A top-heavy plan must provide a minimum contribution to every non-key employee who is a participant and employed at the end of the plan year. Under section 416(c)(2), the minimum is 3% of compensation, or, if lower, the highest contribution rate any key employee received for the year. Matching contributions can count toward it. Top-heavy plans also face faster minimum vesting requirements.

Small, owner-heavy businesses become top-heavy easily, often just because the owners have been in the plan longest and have the largest balances. Safe harbor plans that contain only safe harbor contributions are exempt from the top-heavy rules under section 416(g)(4)(H), but adding any other employer contribution can bring them back in.

Other Tests Run at the Same Time

The ADP, ACP and top-heavy tests are not the only annual checks:

Coverage under section 410(b). The plan must benefit a sufficient share of NHCEs relative to HCEs. The most common measure is the ratio percentage test, which compares the percentage of NHCEs benefiting with the percentage of HCEs benefiting.
Annual additions under section 415(c). Total contributions to a participant's account for the year cannot exceed the lesser of 100% of compensation or the indexed dollar limit, which is $72,000 for 2026.
Compensation limit under section 401(a)(17). Compensation above the indexed limit, $360,000 for 2026, cannot be taken into account in calculating contributions or running tests.
Deferral limit under section 402(g). Each employee's elective deferrals are capped at the annual limit, $24,500 for 2026, plus any catch-up the employee is eligible for. The current retirement plan contribution limits post lists the full set.

The Census: What Payroll Supplies

The testing census is the most important deliverable payroll sends to the plan's administrator each year. A typical census includes, for every employee of every entity in the controlled group, not only participants:

Name, Social Security number and employee ID
Date of birth, date of hire, date of termination and any rehire dates
Hours of service for the year, for eligibility and vesting
Compensation as defined by the plan document, not just gross wages
Pre-tax deferrals, Roth deferrals and catch-up amounts
Employer matching and other employer contributions
Ownership percentages and family relationships to owners, supplied by the sponsor
Officer status
Employment status codes such as union, leased, nonresident alien or excluded class

Why Failures Are Usually Census Errors

When a test fails unexpectedly, look at the data before you look at the participants. The common causes are:

Wrong compensation definition. The plan excludes bonuses but the census reports W-2 Box 1 wages, or the plan uses total compensation but payroll reports base pay. Every employee's ratio is wrong.
Missing eligible employees. Employees who should have been eligible were left off the census, or were never enrolled. Their 0% rate still belongs in the NHCE average, and the failure to enroll them is a separate operational error.
Missing terminated employees. Employees who left during the year still count if they were eligible.
Incorrect hire dates. Rehires coded with a new hire date can wrongly appear ineligible.
Owners' family members not flagged. A spouse working in the business treated as an NHCE distorts both averages.
Related entities omitted. A second company owned by the same owners, with its own payroll, is left out.
Deferrals reported by check date vs. plan year. Payroll at year-end that straddles two plan years needs a consistent cut-off.

Reconciling the census to the year-end payroll registers and Forms W-2 before it goes out catches most of these. Our year-end reconciliation guide covers the payroll side of that tie-out, and the payroll and HRIS integration guide explains where hire and status data typically break between systems.

Correcting a Failed ADP or ACP Test

When a test fails after the data is confirmed, the plan has several correction methods, depending on its document:

Refund excess contributions to HCEs. The excess is calculated by reducing the highest HCE deferral percentages first, then distributed to HCEs based on the highest deferral dollar amounts, as section 401(k)(8)(C) requires. Refunds include allocable income and are taxable to the HCE in the year distributed.
Recharacterize as catch-up. For HCEs who are catch-up eligible and have unused catch-up room, the excess can be treated as a catch-up contribution and stay in the plan.
Qualified nonelective contributions (QNECs) or qualified matching contributions (QMACs). The employer contributes fully vested money to NHCEs to raise their average. This keeps HCE money in the plan but costs the employer.
Forfeiture of unvested match tied to refunded deferrals, for the ACP test.

The Two Deadlines

Deadline	Consequence of missing it
2½ months after plan year end (6 months for plans with an eligible automatic contribution arrangement)	Employer owes a 10% excise tax on the excess contributions under Code section 4979
End of the following plan year	Plan's cash or deferred arrangement can lose its qualified status unless corrected under the IRS's Employee Plans Compliance Resolution System

For a calendar-year plan, March 15 is the date that matters for the excise tax, and December 31 of the following year is the outer limit for corrective refunds or QNECs. Running the test early, ideally in January and February, leaves room to choose the cheapest correction rather than the fastest.

SECURE 2.0 Changes That Affect Testing

Long-term part-time employees. SECURE 2.0 and the original SECURE Act require plans to let certain long-term part-time employees make deferrals. Under section 401(k)(15)(B), employees who are eligible only because of the long-term part-time rule can be excluded from nondiscrimination testing and top-heavy testing. Payroll must track hours carefully to identify these employees in the census.

Roth catch-up. For 2026, participants whose 2025 Social Security wages from the employer exceeded $150,000 must make any catch-up contributions as Roth. When a refund of ADP excess is recharacterized as catch-up for an affected HCE, the Roth rule applies to it as well. The 401(k) Plan Administration handbook covers the broader payroll impact.

How to Make Testing Boring

Confirm the plan's compensation definition and set up a payroll earnings code map that matches it.
Maintain a list of owners, officers and owners' family members, updated annually by the sponsor.
Include every employee of every related entity in the census.
Reconcile census compensation and deferrals to W-2s before sending.
Ask the TPA for a mid-year test projection when HCEs defer heavily.
Calendar March 15 and December 31 of the following year.
Consider a safe harbor design if the plan fails year after year.

Executives' deferred compensation and equity pay add another layer to the compensation definition question. Our guide to executive compensation taxation covers the payroll treatment of those items.

Frequently Asked Questions

What is nondiscrimination testing for a 401(k)?

Nondiscrimination testing is the set of annual tests a 401(k) plan must pass to show it does not favor highly compensated employees. The main tests are the ADP test on employee deferrals, the ACP test on matching and after-tax contributions, the top-heavy test on account balances, and the coverage test on who benefits. The plan's TPA or recordkeeper usually runs them, using census data supplied by payroll and the plan sponsor. Safe harbor plans are treated as passing the ADP test and, if properly designed, the ACP test.

Who is a highly compensated employee for 401(k) testing?

Under Code section 414(q), an employee is an HCE if they owned more than 5% of the employer at any time in the current or prior year, or if their compensation in the prior year exceeded an indexed threshold that the IRS publishes annually. The employer can elect to limit the compensation group to the top 20% of earners. Family members of a 5% owner are treated as owners under attribution rules, so they are HCEs regardless of their pay.

What happens if a 401(k) fails the ADP test?

The plan must correct it, usually by refunding the excess deferrals to HCEs with earnings, recharacterizing the excess as catch-up contributions for eligible HCEs, or making qualified nonelective contributions to NHCEs. If refunds are made within 2½ months after the plan year ends, there is no excise tax. After that, the employer owes a 10% excise tax. Correction must be completed by the end of the following plan year to avoid a qualification problem.

What is the top-heavy test?

A plan is top-heavy if, on the determination date, generally the last day of the prior plan year, key employees hold more than 60% of total plan account balances. Key employees are certain officers above an indexed pay threshold, 5% owners and 1% owners earning over $150,000. A top-heavy plan must provide a minimum contribution, generally 3% of pay or the highest key employee rate if lower, to non-key employees, and must meet faster vesting rules.

Why do 401(k) plans fail nondiscrimination testing?

The plan design or workforce can cause genuine failures, typically when HCEs defer heavily and NHCEs defer little. But many failures come from census errors: compensation reported under the wrong definition, eligible or terminated employees left out, owners' family members not identified as HCEs, or related companies omitted. Reconciling the census to payroll and W-2 data before testing prevents most of these, and often turns a failure into a pass.

What is the difference between an HCE and a key employee?

They are separate definitions used for different tests. HCE status, under section 414(q), drives the ADP and ACP tests and is based on 5% ownership or prior-year compensation above an indexed threshold. Key employee status, under section 416(i), drives the top-heavy test and covers certain officers above a separate indexed threshold, 5% owners and 1% owners earning over $150,000. An employee can be one without being the other.

Own the Data Behind the Test

The TPA runs the math, but the result depends on what payroll sends. The 401(k) Training and Certification Program is the structured route for payroll and benefits staff who want to understand the plan rules behind the census, and the 401(k) Procedures Manual and e-Alerts adds a procedures reference for the annual cycle.