Nineteen states reported 1,069,158 fewer children receiving Supplemental Nutrition Assistance Program benefits than in July 2025. Their all-age SNAP rolls fell by 2,442,853 people over the same state-by-state comparison, according to the Center on Budget and Policy Priorities. [2]

That measured contraction advances this paper's July 9 position on SNAP cuts arriving as the government canceled the comparable hunger survey. The earlier article said participation loss could not by itself prove a hunger outcome. Position 2 now supplies a child-participation table, but the missing national survey still prevents the table from answering what happened inside households.

The Guardian leads with more than 1 million children losing food aid and reports CBPP's estimate that the national child decline exceeds 1.5 million. [1] Those are not the same number. The 1,069,158 count comes from 19 states with child data. The larger figure is CBPP's conservative estimate for the country, produced by assigning 30 percent of the all-age decline to children in states without child counts. [2]

That method is an effort to describe a missing national ledger, not a USDA national child series. Adding the confirmed 19-state count to the estimate would count the same problem twice. Treating the estimate as an agency measurement would give it an authority it does not claim.

Nineteen states and many dates

The combined state record shows a 12 percent decline among participants of all ages and a 13 percent decline among children. Children accounted for 44 percent of the combined drop. [2] The precision of those totals should not disguise the uneven calendar beneath them.

Vermont's latest comparison runs through March 2026. New Jersey and New Mexico run through April. Several states use May, while others use June. Alabama, Arkansas, California, New Hampshire, Pennsylvania, and Vermont supplied unpublished state-agency data to CBPP. Kentucky was omitted because of data-quality concerns. [2]

The state variation is large. Arizona reported 182,058 fewer children, a 49 percent decline. Texas reported 317,316 fewer, down 18 percent. California reported 156,440 fewer, down 9 percent. New Hampshire's child decline was 945, or 4 percent. [2] Those figures show why one national verb cannot explain every state.

Definitions also matter. Ohio defines children in this table as people under 22 who receive SNAP and live with a parent. [2] A child count built from administrative systems is only as comparable as the age, household, and reporting rules beneath it. The table is strong evidence of participation change, not a synchronized census taken under one clock.

CBPP says the primary drivers cannot yet be isolated. Eligibility changes, renewals, administrative friction, voluntary exits, migration, and later data revisions can all affect participation. A removed case is not automatically a family whose income improved. It is also not automatically proof that a child missed a meal.

Policy and consequence remain separate

The 2025 reconciliation law is enacted. CBPP links part of the access problem to a future state cost shift scheduled to begin in October 2027. [2] That schedule can influence planning and political choices now, but state payments under it have not begun. Further restrictions, state withdrawals, or a congressional delay remain contingent.

USDA's eligibility guidance remains a mutable federal reference for who can receive SNAP under program rules. [3] It does not supply the missing comparable national child-participation series, and a later page update must not silently rewrite the state snapshot used here. Eligibility, successful renewal, actual participation, benefit receipt, and food security are separate stages.

The canceled hunger survey makes that separation more consequential. Participation answers whether a person appears on the program roll. Food security asks whether a household can reliably obtain enough food. The earlier article warned that ending the long-running common measure would make any policy result harder to judge. The new child table confirms contraction without restoring the outcome test.

That evidentiary gap constrains both political camps. A smaller roll cannot prove that fraud or ineligible participation accounted for the decline. A large child decline cannot, by itself, produce a measured national increase in hunger. Food banks, school systems, state surveys, and local researchers may observe consequences, but unlike records cannot simply be assembled into the canceled federal series.

CBPP's official X post calls the decline an "unfolding emergency" and urges Congress to delay the state cost shift. It links the exact analysis behind this article and supplies an advocacy frame, not a second measurement. The post cannot turn participation data into a hunger outcome or isolate why children left the rolls.

Mainstream loss language has an understandable force because 1,069,158 children disappearing from 19 state rolls is a large completed administrative fact. [1] [2] Yet lost SNAP still names a benefit status, not a cause or household outcome. The paper's responsibility is to publish the contraction without giving the table work it cannot do.

The next useful record should be more granular, not louder. USDA or states could publish compatible national child counts. States could explain renewals, denials, notices, processing times, and data revisions. Congress could decide whether the October 2027 cost shift remains. A restored comparable hunger survey could finally reconnect participation changes to the condition the program is designed to address.

Until then, the evidence supports two statements at once. Nineteen states recorded more than 1 million fewer children on SNAP. The country still lacks the common federal instrument needed to say what that contraction did to child hunger. The first number is now visible. The consequence remains hidden by policy choice.

-- NORA WHITFIELD, Chicago