Foreign-born share of county population, 1990–2024
Share of each county’s residents born outside the US (excluding people born in Puerto Rico, US island areas, or abroad to American parents). Census 1990 and 2000 long-form sample, then American Community Survey 5-year estimates. Every frame is mapped onto 2023 county boundaries.
National totals by vintage (table view)
How to read this / caveats
Vintages are not single years. “ACS 2019–2023 5-yr” pools five years of survey responses; it is the only ACS product that covers every county. Consecutive ACS frames overlap by four years, so they move smoothly by construction. Census 1990 and 2000 values are from the decennial long-form sample (about 1 in 6 households).
Small counties are noisy. Where the ACS margin of error (90%) on the foreign-born count exceeds 30% of the estimate, the tooltip says so and “Mark low-reliability estimates” hatches the county. About four in ten counties are flagged in a typical frame, together holding about 5% of the population.
Boundaries. Connecticut’s 2022 planning regions are rebuilt exactly from town-level data for 2000–2021 and approximated for 1990. Former Virginia independent cities (Bedford, Clifton Forge, South Boston) are merged into their surrounding counties. Some Alaska census areas that were later split share one rate across their successors. Broomfield County, CO did not exist before 2001 and is blank for 1990 and 2000.
Group quarters. Census and ACS count people where they live on the reference date, including prisons, immigration detention centers, college dorms, military barracks and nursing homes. Counties where institutional, college and military group quarters together hold 5% or more of the 2020 Census population are flagged with the dominant type (2020 Census PL 94-171 table P5; ICE detention is tabulated as adult correctional). In those counties a jump in the foreign-born count can reflect a facility rather than a community.
Country of birth. Some countries are only tabulated separately from a later vintage (Nepal, Burma, Yemen and Eritrea from ACS 2005–09; Somalia from ACS 2011–15; DR Congo from ACS 2015–19; Bhutan and the Marshall Islands from ACS 2018–22). Earlier vintages show no data and the baseline moves to the first published vintage.
Citizenship and year of entry. ACS table B05001 (Census 2000 SF3 P021) splits the foreign-born into naturalized citizens and noncitizens; B05005 (SF3 P022) gives the year each person came to live in the US. “Noncitizen” is not a legal-status category: it lumps together green-card holders (lawful permanent residents), temporary visa holders such as students and workers, refugees and asylees, and unauthorized immigrants. The Census does not ask legal status, so no layer here can show how many residents are unauthorized. Year of entry is the most recent arrival, and “entered 2010 or later” is an open-ended window, so later vintages include more arrival years by construction.
Administrative records and modeled estimates. Record layers (refugee arrivals, sponsor releases, green cards, labor certifications, court cases) count events per year: a flow, not the people living in a county. They are not comparable with the ACS foreign-born count or with each other, and must not be added up. Where a source reports by city or metro area, values are allocated to counties and hatched. Modeled estimates are labeled as such, with the modeler and year, and show their range where one is published. Each layer’s panel lists what one unit counts, its caveats and its sources.
Community pins mark case-study places with a documented arrival history; each pin appears from its arrival start year onward. Program/pathway attribution and sources are in each pin’s panel, graded A (primary), B (solid reporting), C (thin).