Methodology
Last updated · Source data: plan year 2024 Form 5500 filings
Rankings derived from U.S. Department of Labor Form 5500 filings, plan year 2024.
Where does this data come from?
Every number on this site comes from the U.S. Department of Labor’s public Form 5500 bulk data for plan year 2024 — specifically Schedule H, where each large plan names its auditor, the auditor’s EIN, and the opinion issued. No firm pays to appear or rank.
How the rankings are built
- Download. We take the DOL’s EFAST2 bulk CSV files for the latest complete plan year (2024): the Form 5500 header file and the Schedule H file.
- Identify audits. Every Schedule H row naming an accountant firm is an audited filing. We capture the firm’s name, EIN, and opinion type as filed.
- Classify plans. Joining to the Form 5500 header, we keep retirement plans (single-, multi-, and multiple-employer plans with a pension benefit code) separate from large welfare plans and direct filing entities.
- Group by firm. Filings are aggregated by auditor EIN; the display name is the most common spelling across the firm’s filings. Slugs, ordering, and tie-breaks are deterministic, so a rebuild on the same inputs is byte-identical.
- Publish. National, state, and firm pages are regenerated as static HTML from the resulting dataset. Nothing is editorially reordered.
Definitions
| Term | Definition |
|---|---|
| Audited filing | A Form 5500 filing whose Schedule H names an independent qualified public accountant (IQPA) — i.e., a plan large enough to require an audit. |
| Retirement plan | A filing by a single-employer, multiemployer, or multiple-employer plan (entity codes 1–3) carrying a pension benefit code: 401(k), 403(b), profit-sharing, ESOP, or defined benefit. Excludes welfare-only filings and direct filing entities (DFEs). |
| Audited filings (league-table basis) | All audited filings signed by the firm, including large welfare plans. This is the basis of the national top-100 table and matches published league tables. |
| Retirement plans audited | The subset of a firm’s audited filings that are retirement plans. Used for state rankings, opinion mix, client sizes, and per-firm profiles. |
| Firm | All filings sharing an accountant EIN, displayed under the most common spelling of the firm’s name in the filings. |
| Participants | Total participant count at the beginning of the plan year as filed (falling back to active participants when blank). |
| State | The plan sponsor’s address state on Form 5500 — where the client is, not where the audit firm’s offices are. |
| Opinion type | Schedule H accountant’s opinion code: unmodified, qualified, disclaimer, or adverse. |
Two counting bases appear on the site, always labeled: the national top-100 uses all audited filings including welfare plans (league-table basis), while state pages, firm opinion mixes, and client-size figures use retirement plan filings only. Plan year 2024 totals: 77,427 audited filings, 72,776 of them retirement plans, audited by 4,052 firms.
Reading the opinion tables
Opinion type is transcribed from Schedule H exactly as filed. One nuance prevents misreading: for decades, most plans elected the “limited-scope” audit, under which auditors issued a disclaimer of opinion as a matter of form. AICPA SAS 136 replaced that format with the ERISA Section 103(a)(3)(C) report, but disclaimers still appear in filings. A disclaimer in these tables is therefore usually a legacy reporting convention or coding choice — not evidence of a failed audit. Adverse and qualified opinions are the categories that warrant follow-up.
Known limitations
- EIN-level grouping. Firms that merge, rebrand, or file under multiple EINs (common after CPA-firm private-equity deals) can appear split or combined until filings catch up.
- Sponsor-entered data. Participant counts, names, and EINs are transcribed from filings and inherit any filer errors.
- Timing. A plan year’s data stabilizes as extended filings arrive; late and amended filings after our snapshot are not reflected until the next refresh.
- No fee data. Form 5500 does not reliably report audit fees, so cost figures on this site are labeled editorial estimates and are never attributed to the dataset.
- Volume is not quality. Engagement counts measure experience, not audit quality; we publish opinion mix and DOL research as context, not verdicts.
Updates and corrections
The dataset is refreshed when the Department of Labor publishes each plan year’s bulk files, and the “Last updated” date on every page reflects the current build. Firms and sponsors who spot an error can request a correction via the about page; we correct against the underlying filings, not on assertion.
Every refresh is logged with its source-file dates and record counts on the data updates page.
Source: U.S. Department of Labor, EFAST2 Form 5500 datasets, dol.gov Form 5500 datasets.