The Section 25F Safe Harbor: Why the 90% Test Nearly Broke Every Diversified Nonprofit — and What Changed in June

Section 25F requires a Scholarship Granting Organization to spend not less than 90% of its income on scholarships for eligible students. Every SGO operator knows the number. What most organizations have not fully absorbed is that the definition of "income" in that sentence was, until June, the single largest structural threat in the program — and that Treasury's June 9, 2026 guidance preview changed it in a way that should determine how new SGOs are structured.

The Problem: What Counts as "Income"?

The statute says 90% of "income of the organization" must go to scholarships. It does not define the term. IRS Notice 2025-70, issued in late 2025, described Treasury's anticipated approach: income would include all income of the organization — including unrelated business income — and would not be limited to the qualified contributions held in the segregated scholarship account.

Consider what that means for any organization that does more than grant scholarships. A school association with membership dues, accreditation fees, curriculum sales, and events revenue. A community foundation with management fees and multiple program areas. A church network with congregational support. Under the Notice's reading, if such an organization became an SGO, it would need to spend 90% of its entire organizational income — every dues dollar, every program fee — on K-12 scholarships.

That is not a compliance burden. It is a structural impossibility. Under the general rule as previewed, the 90% requirement is measured against total receipts, unreduced by expenses — an organization cannot even net out the cost of its other programs before the test applies. The literal Notice position made it flatly irrational for any diversified organization to seek SGO status. The only viable path was a stripped-down entity that did nothing else.

The Fix: The June Safe Harbor

Treasury's June preview introduced the safe harbor that resolves this. If an organization's activities are largely scholarship-granting, it may measure income for the 90% test by the amount held in its Section 25F segregated account — the qualified contributions themselves, plus earnings on them.

Under the safe harbor, the test becomes what most people assumed it always was: of the money donors contributed for scholarships, at least 90% must be spent on scholarships. Money that never enters the segregated account — dues, grants for operations, general support gifts — is simply not in the denominator.

For a multistate SGO, the safe harbor must be satisfied separately for each state-specific segregated account. There is no blending across states — a point with real consequences that we cover in our multistate structure analysis.

The Condition: "Largely Scholarship-Granting"

The safe harbor is conditional, and the condition is the part every board should read twice. It is available to organizations whose activities are largely scholarship-granting.

Treasury has not defined "largely." There is no percentage, no revenue threshold, no activity test — nothing yet that tells an organization how much non-scholarship activity forfeits the safe harbor and throws it back onto the general rule, where the test runs against total receipts. That question is on the open list for the September proposed regulations.

What is already clear is the direction: a diversified organization, as a whole, does not qualify. An entity whose overwhelming purpose and activity is granting scholarships does.

The Structural Consequence: Form a Separate Entity

Put the pieces together and the design answer writes itself. An organization with existing programs and revenue should not itself become the SGO. It should form a separate affiliated entity — its own 501(c)(3), not a private foundation — whose activities are largely scholarship-granting, and let that entity seek listing.

The affiliated-entity structure does three things at once:

• It secures the safe harbor. The scholarship entity's activities are, by construction, largely scholarship-granting. The 90% test runs against its segregated accounts, not against the parent organization's budget.
• It quarantines the general-rule risk. If the September regulations draw the "largely scholarship-granting" line somewhere unexpected, a dedicated entity is on the right side of any plausible line. A diversified organization is betting its whole budget on a definition that does not exist yet.
• It keeps the parent's operations out of scope. The parent's dues, programs, and reserves stay outside the SGO's compliance perimeter — while remaining available, if the parent chooses, to fund the SGO's operations as money that never touches the segregated accounts and never enters the denominator.

That last point has a corollary worth naming: under the safe harbor, an SGO's overhead does not have to be funded from the 10% at all. General operating gifts — deductible under the ordinary charitable rules rather than credited under Section 25F — sit outside the test entirely. The two-gift structure this enables is a topic for its own post, but the safe harbor is what makes it work.

The Caveats

Three cautions before anyone treats this as settled.

It is a preview, not a regulation. Treasury has said proposed regulations will arrive by the end of September 2026 and will be reliable for tax year 2027. Until then, the safe harbor's precise contours — especially the "largely scholarship-granting" condition — can move.

The 90% is a spending test, not just a ratio. Money sitting undisbursed in a segregated account is its own compliance problem. The safe harbor defines the denominator; it does not relax the obligation to actually move scholarship dollars to students.

Year one remains hard. A new SGO's launch costs are front-loaded, and a first partial year of contributions cannot absorb them inside a 10% allowance. Notice 2025-70 explicitly asked whether the regulations should address start-up costs or multi-year smoothing. No answer has been given. Prudent modeling assumes no relief and treats any September accommodation as upside — which means year-one funding must come from outside the segregated accounts.

The Bottom Line

The June safe harbor converted the 90% test from an existential threat into an engineering constraint. But it rewards one structure heavily over all others: a dedicated, largely-scholarship-granting entity with clean segregated accounts, operating alongside — not inside — whatever organization gave rise to it. Organizations making formation decisions this summer should make them with that structure as the default, and should have exempt-organizations counsel pressure-test anything else.

For the full context of the June preview — the audit rules, the unique donor number, income verification, and the rest — see our complete item-by-item walkthrough.