IRS SOI Statistics: S Corporations and Partnerships

Why entity type matters for deal screening

A lower-middle-market target list is mostly a list of closely held businesses, and closely held businesses file taxes in a few predictable forms: S corporations on Form 1120-S, partnerships on Form 1065, other corporations on the Form 1120 series, and sole proprietors on Schedule C ^[1]. Most partnerships are LLCs ^[2]. The form matters to a buyer because it shapes deal structure and the seller's tax position.

Census business counts, covered in our comparison of CBP vs QCEW vs SUSB, count firms and establishments by size and industry. They do not split the universe by tax form. The IRS Statistics of Income (SOI) division does. This article maps what SOI publishes on S corporations, partnerships and corporations, gives the latest counts verified on IRS pages on 2026-10-28, and shows how to use the tables to bracket a target universe. It describes published statistics, not tax rules for any particular business, and is not tax advice.

Two IRS definitions frame the S corporation numbers. To qualify, a corporation must be domestic, have only allowable shareholders (individuals, certain trusts and estates, and not partnerships, corporations or non-resident aliens), have no more than 100 shareholders, have one class of stock, and not be an ineligible corporation ^[3]. S corporations "elect to pass corporate income, losses, deductions, and credits through to their shareholders for federal tax purposes" ^[3].

What IRS SOI publishes

SOI's business tax statistics page groups its studies by subject. The ones that matter for sizing an owner-operated universe are ^[1]:

Source: IRS SOI, Business tax statistics.

One navigation trap is worth knowing. The SOI S corporation page links basic Form 1120-S tables only through tax year 2017 ^[4]. Its featured article reports tax year 2003 ^[4]. The current Form 1120-S tables, through tax year 2023, are published as part of the Corporation Complete Report, Publication 16 ^[5]. Analysts who stop at the S corporation page will quote stale numbers.

Partnerships: the latest verified counts

The SOI partnership page reports, for tax year 2023 ^[2]:

• Partnerships "filed over 4.5 million returns for TY 2023, a 1.7% increase over the number filed for 2022."
• Those returns "represented more than 30.2 million partners, a 5.0% increase from the previous year."
• Limited liability companies "made up most partnerships (72.7%)."
• Limited partnerships were 9.7% of partnerships but reported 32.8% of pass-through income and 34.2% of partners.
• Real estate and rental and leasing accounted for 50.7% of partnerships; finance and insurance reported the largest shares of pass-through income (63.5%), total assets (59.9%) and total receipts (25.0%).

The sector split is the first thing to notice when sizing. Half of all partnership returns are real estate, so a raw partnership count says little about operating businesses. Filter by sector before using it. Tiering is the second trap: partners that are themselves partnerships received the largest share of income allocated to partners, $635.5 billion ^[2]. Some partnership returns are holding or fund layers above an operating business, not separate businesses. The page offers tables by sector or industry, by entity type and by size of total assets ^[2].

S corporations and corporations: what the TY2023 complete report shows

Publication 16 (Rev. 9-2026) covers tax year 2023. It presents "statistical estimates based on a stratified sample of 129,240 unaudited returns that were selected from the approximate 7 million active corporate returns that were filed for Tax Year (TY) 2023" ^[6]. Of those, "approximately 5.5 million were passthrough entities, which include RICs, REITs, and S corporations" ^[6]. For 2023, "S corporations reported $814.5 billion in pretax profits, an increase of 6.7% from 2022" ^[6].

The exact counts are in the Publication 16 tables:

All three are sample-based estimates ^[7][8]. Dividing the two return counts, Form 1120-S returns were about 77.4% of active corporation returns in TY2023. That is S corporations as a share of corporations, not of all businesses: it leaves out partnerships and sole proprietorships, which SOI tables separately.

The shareholder count is a useful signal for owner-operator work. Dividing Table 3.2's shareholders by its returns gives about 1.5 shareholders per Form 1120-S return on average ^[8]. SOI also publishes Table 9, which classifies Form 1120-S returns by number of shareholders ^[5], if a thesis depends on closely held ownership.

For history, the S corporation page's featured article reported that for tax year 2003 "about 61.9 percent of all corporations filed a Form 1120S," nearly 3.3 million returns, and that S corporations "became the most common corporate entity type in 1997" ^[4].

The size band that matters for sponsors

Count by receipts, not by all returns. Table 3.1 (all active corporations) and Table 3.2 (Form 1120-S) use the same business-receipts size classes for TY2023:

Sources: Table 3.1 and Table 3.2. The all-corporations figure for $50 million to $250 million adds Table 3.1's two classes, $50,000,000 under $100,000,000 (25,459) and $100,000,000 under $250,000,000 (15,364). Shares are computed from the table values.

The pattern is the useful part: in these tables the S corporation share falls as receipts rise, from about 73% in the $5 million to $10 million band to about 55% at $50 million to $250 million. A thesis aimed at companies with $10 million to $50 million in receipts should expect roughly two of every three corporate targets in that band to file as S corporations, before any industry cut. Table notes add one wrinkle: for "Finance and Insurance" and "Management of Companies (Holding Companies)," SOI classifies by total receipts instead of business receipts ^[8].

Integrated business data

The Integrated Business Dataset (IBD) is the one SOI product that puts corporations, partnerships and sole proprietorships side by side. It "was assembled at the table level from the annual SOI cross-sectional studies of corporations (C and S corporations), partnerships, and nonfarm sole proprietorships," and data from exempt organizations with unrelated business income (Form 990-T) "are not included" ^[9].

It is also stale. The headline table of returns, receipts and net income by form of business covers 1980-2015, and the sector table's latest year is 2015 ^[9]. Use it for long-run history of business form, not for a current count. For a current cross-form picture you have to read the corporation, partnership and sole proprietorship releases separately and keep their definitions straight. This article does not add them into a single "pass-through share," because the forms are counted on different bases.

The sole proprietorship release shows why. SOI's Table 1 for nonfarm sole proprietorships counts 31,125,909 returns for tax year 2023 across all nonfarm industries ^[10], more than five times the Form 1120-S count. Most of that volume is far below any acquisition threshold, so it belongs in a sizing exercise only after a receipts filter (SOI's size-of-receipts tables for sole proprietorships currently run through tax year 2020) ^[11].

Using SOI to bracket a target universe

SOI cannot produce a target list, but it can tell you whether a list is complete. The method is to build a top-down bracket from SOI and Census, then compare it with your bottom-up list.

1. Pick the size measure. Receipts bands (Tables 3.1 and 3.2) map more naturally to a revenue thesis than asset bands (Tables 2.1 and 2.4) ^[5].
2. Pick the industry cut. The Form 1120-S tables are classified either by size or by industry, not usually both: Table 2.4A crosses sector with asset size; Tables 6.1 and 7 classify by major industry; Table 3.2 classifies by receipts size ^[5]. Expect to combine a size share and an industry share, and label the result an estimate.
3. Add partnerships separately. Use the partnership tables by sector and asset size ^[2]. Do not add them to corporate counts in the same band without checking that the size measure matches.
4. Cross-check against Census. Census SUSB counts employer firms by size and industry. If SOI implies many more filers than SUSB has employer firms in the same cut, part of the gap is likely filers with no employees, holding entities, or definitional differences, not missing targets.
5. Compare with the list. A list far below the bracket is thin; a list above its high end likely contains duplicates or out-of-scope records.

Illustrative example (fictional)

A thesis targets "regional commercial cleaning companies" with $10 million to $50 million in receipts. The analyst starts from the real TY2023 figure of 119,562 Form 1120-S returns in that receipts band ^[8], then applies an assumed industry share. Everything below the first line is fictional and only shows the arithmetic.

The output is a range to argue about, not a fact. The industry share is the weakest input and should come from an industry table, not a guess. The point is the order of magnitude: a list of 610 against an implied universe near 1,400 says the list is incomplete. For the wider workflow, see Docket's notes on market mapping.

Caveats

• Tax-year lag. TY2023 "includes accounting periods ending July 2023 through June 2024" ^[6], and the tables carry the note "2023 Publication 16, September 2026" ^[8]. In late 2026 the newest corporate data describes businesses as they were two to three years earlier.
• Estimates, not counts. Tables state "All figures are estimates based on samples," flag some cells with an asterisk to be "used with caution because of the small number of sample returns," and delete some cells "to avoid disclosure of information for specific corporations" ^[8]. Publication 16 also publishes coefficients of variation for Table 1 ^[5].
• Returns, not firms or establishments. A consolidated return "filed by the common parent company was treated as a unit" ^[6], so a group with many subsidiaries may count once. One business can also operate many establishments. SOI counts tax filers; Census counts firms and establishments. See CBP vs QCEW vs SUSB for the Census side.
• Active returns only. The statistics "include active domestic and foreign corporations and exclude inactive corporations," and include part-year returns ^[6].
• No names. SOI tables are aggregates. They cannot tell you which companies are in a band, only how many.

Worksheet

Copy this into a sheet and fill one row per figure you plan to quote.

Docket's Scout and Audit agents work the bottom-up side of this comparison: each recorded answer about a target quotes a retained page, and anything that cannot be quoted stays open.

Sources

1. SOI Tax Stats - Business tax statistics. Internal Revenue Service. Accessed 2026-10-28.
2. SOI Tax Stats - Partnership statistics. Internal Revenue Service. Accessed 2026-10-28.
3. SOI Tax Stats - S corporation statistics. Internal Revenue Service. Accessed 2026-10-28.
4. SOI Tax Stats - Corporation Income Tax Returns Complete Report (Publication 16). Internal Revenue Service. Accessed 2026-10-28.
5. Publication 16, Corporation Income Tax Returns Complete Report, Tax Year 2023 (Rev. 9-2026). Internal Revenue Service, Statistics of Income Division, September 2026. Accessed 2026-10-28.
6. Returns of Active Corporations, Table 2.1, Tax Year 2023 (XLSX). IRS Statistics of Income Division, September 2026. Accessed 2026-10-28.
7. Returns of Active Corporations, Table 3.1, Tax Year 2023 (XLSX). IRS Statistics of Income Division, September 2026. Accessed 2026-10-28.
8. Returns of Active Corporations, Form 1120-S, Table 3.2, Tax Year 2023 (XLSX). IRS Statistics of Income Division, September 2026. Accessed 2026-10-28.
9. SOI Tax Stats - Integrated business data. Internal Revenue Service. Accessed 2026-10-28.
10. S corporations. Internal Revenue Service. Accessed 2026-10-28.
11. SOI Tax Stats - Nonfarm sole proprietorship statistics. Internal Revenue Service. Accessed 2026-10-28.
12. Nonfarm Sole Proprietorships, Table 1, Tax Year 2023 (XLS). IRS Statistics of Income Division, Sole Proprietorship Returns 2023, March 2026. Accessed 2026-10-28.