A PDF of this resource is available here.
The Trump administration has proposed changes to the 2030 census that would exclude many noncitizens from the population count used to allocate seats in the U.S. House of Representatives, prohibit certain demographic questions, and change how the Census Bureau determines where people live. The administration is also considering how to identify respondents’ citizenship and immigration status, potentially through government records or survey questions. The proposed rule on the changes permits public comments until November 2, 2026. If finalized, the final rule including the changes would likely face legal challenges.
What would change?
The Census Bureau’s proposal has two main components: 1) Revised residence criteria governing who is counted and where, and 2) restrictions on demographic questions. The administration says these changes would better implement its interpretation of the U.S. Constitution, protect census data, and promote efficiency. Unlike the residence criteria issued as departmental policy in advance of the 2020 census, the new proposed criteria would be adopted using formal rulemaking and incorporated into the Code of Federal Regulations.
Excluding many noncitizens from congressional apportionment
As part of the new residence criteria, the proposal would change the meaning of “usual residence” for noncitizens, with an aim toward excluding large numbers of noncitizens from being counted for population and apportionment[1] purposes. In a departure from longstanding policy and practice, the Census Bureau would treat immigration status and what it describes as an “enduring tie” to the United States as conditions for inclusion in the apportionment count, rather than simply relying primarily on where someone lives. As a result, many noncitizens who live and work in U.S. communities would be excluded from being counted even if their U.S. address is undisputed. The bureau is considering using government records, including immigration records, and potentially questions on citizenship and immigration status to determine who qualifies for exclusion under this new rule. A previous effort to exclude undocumented immigrants from the 2020 apportionment count occurred during President Trump’s first term, but that policy was not implemented before President Trump left office. Critics have characterized the new standard as unconstitutional, ahistoric, and harmful.
The census has historically counted people at their usual U.S. residence regardless of citizenship or immigration status, while excluding short-term visitors such as tourists. Section 2 of the Fourteenth Amendment requires congressional seats to be apportioned among the states by “counting the whole number of persons in each State” (emphasis added), rather than specifying that only citizens ought to be counted. The Census Bureau’s historical account states that the apportionment population “always has included those persons who have established a residence in the United States,” and that this principle dates to the first Census Act in 1790. Historically, that population has included both citizens and noncitizens, regardless of immigration status.
The proposed rule would extend beyond excluding undocumented immigrants and would also exclude many categories of legal residents that have been counted previously. Under the proposed rule, foreign nationals would be excluded from the congressional apportionment count unless they are also U.S. citizens (dual nationals) or lawful permanent residents (commonly known as green card holders). In addition to excluding the undocumented population, which historically has always been included in the count, several categories of residents with legal presence would also be excluded, including authorized workers on nonimmigrant visas, people with Temporary Protected Status (TPS), Deferred Action for Childhood Arrivals (DACA) recipients, refugees, asylum seekers who have not adjusted to permanent residency, and international students. Estimates suggest the affected population could account for as much as 5% of U.S. residents, representing a significant undercount.
Excluding noncitizen residents would change the population totals used to apportion House seats among states. Because each state’s Electoral College allocation includes its House delegation, changes in House seats would also change its electoral votes. The consequences would depend on the distribution of excluded residents relative to each state’s total population, not simply on which states have the largest immigrant populations. The Forum’s 2018 analysis (which concerned reduced participation caused by a citizenship question rather than explicit exclusions like those in the current proposed rule) warned that an immigrant undercount in the 2020 census could reduce representation not only in Democratic-leaning states, but also in Republican-leaning states such as Texas and Florida. A recent analysis similarly found that Texas, and potentially Florida, may receive fewer congressional seats under the proposal than under a full population count.
Ending collection of certain demographic information
A separate component of the proposed rule would prohibit questions about race, ethnicity, and sexual orientation on the decennial census short-form questionnaire and other questionnaires used to enumerate the population. The administration cites the potential sensitivity of these questions and what it describes as their reduced need or usefulness. The census has collected racial classifications since 1790, although the categories and methods have changed substantially.[2] Today, racial and ethnic data help guide redistricting, enforce voting rights and other antidiscrimination laws, and assess differences in health outcomes and access to services. Civil rights advocates warn that removing these questions would weaken the evidence used to identify discrimination and challenge it in court, including in voting rights cases.
The restrictions concern questionnaires used for the decennial population count, rather than an across-the-board prohibition on demographic questions in every Census Bureau survey. Other surveys, including the American Community Survey (ACS), would remain distinct sources of demographic information. The ACS is an ongoing Census Bureau survey that samples approximately 3.5 million addresses annually and collects information about housing, education, employment, and other characteristics. Unlike the decennial census, it does not attempt to count every resident. Its estimates therefore would not replace the decennial census’s detailed racial and ethnic counts for individual census blocks, which are used in redistricting and voting-rights enforcement.
Census-derived statistics help guide federal funding for health care, education, housing, transportation, and other services. Excluding residents or reducing participation could affect the data used to distribute those resources. The consequences would vary by program, its funding formula, and how the revised population data are incorporated. The effects would reach beyond immigrants themselves. State and local governments also use population data to plan services for entire communities, including schools and emergency response. Undercounts can affect rural and urban communities in both Republican- and Democratic-leaning states.
Potential inclusion of a citizenship and/or immigration status question
The proposed rule does not explicitly call for the creation of a citizenship question or an immigration status question in the 2030 census, but the proposed rule would pave the way for their potential inclusion.[3]
Because the proposed rule would exclude most noncitizens from the census count, it would need to make determinations on individuals’ immigration statuses. While the proposal does not specify exactly how the Census Bureau would make those determinations, it presumably could seek to include a citizenship or immigration status question in the 2030 census in addition to relying on existing government records.
Neither a citizenship question nor an immigration status question, by themselves, exclude someone from the population count, however research indicates these types of questions make individuals less likely to respond to the census. A 2025 peer-reviewed study examining a Census Bureau experiment found that adding a citizenship question would likely reduce self-response among several groups, including households with noncitizens, and increase omissions of household members among those who respond. Citizens and legal residents in mixed-status households may avoid responding out of concern for undocumented relatives. Reduced participation could therefore affect U.S. citizens and green card holders, as well as people the proposal expressly seeks to exclude.
The potential inclusion of citizenship or immigration status question in 2030 follows the first Trump administration’s unsuccessful effort to add a citizenship question to the 2020 census. In June 2019, the Supreme Court blocked the addition of a citizenship question to the 2020 census, finding that its stated justification of improving Voting Rights Act enforcement appeared “contrived.” The Court returned the matter to the Commerce Department for further explanation, and the question did not appear on the final questionnaire. The ruling, however, did not prohibit citizenship questions categorically – it held that the Constitution may permit asking about citizenship but required a genuine explanation for the agency’s decision.
Could courts or the next administration reverse the changes?
If finalized, the rule is expected to face litigation. The administration argues that certain noncitizens lack the “allegiance” or “enduring ties” to the United States necessary to qualify as inhabitants for apportionment purposes. That interpretation departs from longstanding census practice, which has never utilized this newly articulated “allegiance” principle,[4] and the bureau’s own proposal acknowledges that the 2020 residence criteria counted foreign citizens living in the United States without distinguishing between those with and without lawful immigration status. Critics maintain that the proposed exclusions conflict with the Constitution’s use of “persons,” rather than “citizens,” and with the historical inclusion of noncitizen residents.
Aside from litigation, policymakers may attempt to revoke a final rule issued in 2026 prior to the 2030 census. The president taking office in January 2029 will oversee the 2030 census and could seek to revise or rescind such a rule. Congress could also address the changes through legislation. Reversal would not be automatic, especially given the truncated timeframe – amending or repealing a binding regulation generally requires further rulemaking, including notice and comment, and a reasoned explanation for the policy change. Federal law requires the Commerce Secretary to report proposed census subjects and types of information to the relevant congressional committees by April 1, 2027, and proposed questions by April 1, 2028. Those deadlines do not categorically prevent later changes: The law permits modifications before the census begins if the Secretary finds that new circumstances necessitate them and reports the proposed modifications to those committees. The proposal’s ultimate effect will therefore depend on the final rule, litigation, congressional action, and decisions by the next administration.
[1] Apportionment is the process of dividing House seats among the states based on population.
[2] The first census distinguished among free white people, “all other free persons,” and enslaved people; it did not use today’s self-reported race and ethnicity questions.
[3] A citizenship question asks whether each household member is a U.S. citizen; an immigration status question would seek additional information to distinguish among noncitizens.
[4] In a different context, the U.S. Supreme Court rejected the application of an “allegiance” principle in its recent ruling in Trump v. Barbara, upholding birthright citizenship for those born within the territory of the United States, subject to a few narrow exceptions.