Migration data
Tabulation of current state of residence crossed by state of residence 1 year ago. This is a 1-year mobility / previous-residence statistic, not a count of household-goods shipments.
- File
- State-to-State Migration Flows: 2024 American Community Survey 1-Year Estimates (Table T13)
- Period
- 2024
- Official release
- Jan 21, 2026
- Last checked
- Sep 7, 2026
- Status
- latest available
- Universe
- Population 1 year and over
- Retrieved
- Sep 6, 2026
What one cell means
A cell in Table T13 is an estimate of the number of people, aged 1 and over, whose current residence is one state and whose residence one year earlier was another. It says nothing about when in the year they moved, how many times they moved, whether they moved as a household, or whether goods were shipped. YardPermit never describes a cell as a count of moves, shipments or trucks, and the calendar year is not the dataset year: the 2024 table was released in 2026, and no 2025 state-to-state table is published.
Ingestion
The workbook is read once by a script that writes the site’s data files; 2,652 pairwise cells and the national totals were taken from it. Cells the Census Bureau marks as not available are kept as not available and never imputed, averaged or carried over from an earlier vintage. The script does not estimate, smooth, project or round anything beyond what the table already contains.
Margins of error
Published ACS margins of error are 90 percent margins of error. Estimate ± MOE is a 90 percent confidence interval. Standard error ≈ MOE / 1.645. Every estimate on the site is printed with its margin, and pages that compare two estimates test the difference first. The Census Bureau advises that differences between survey estimates be tested for statistical significance rather than read at face value, and its own testing tool works at a standard 90% confidence level. YardPermit applies the standard rule for two independent estimates: the margin of error of a difference is the square root of the sum of the two squared margins, √(MOE₁² + MOE₂²). A net balance or a gap between two flows is reported as established only when it exceeds that margin; otherwise the page says the table does not establish a direction.
Derived figures
Ranks are competition ranks among cells that carry an estimate, so ties share a rank and a cell with no estimate has no rank. A share divides a pair estimate by the state’s published in-from-other-state or out-to-other-state total. A 90% interval is the estimate minus and plus its margin, floored at zero. Net balance is in minus out with the difference test above. Each derivation is printed beside the figure, and the modules that compute them have tests that pin their arithmetic.
On each state page
The written-out version of what each state page links here instead of repeating. The two totals at the top, in from other states and out to other states, are previous-residence counts of people aged 1 and over from Table T13; they are not household-goods shipments or permanent relocations, and moves from abroad are not in them. The two-way exchanges add a pair’s two directions together for the five largest pairs; that sum is not a net gain, a count of unique people or of shipments. Each pair’s direction is tested with the difference rule above, and a direction inside its margin is left unresolved.
The population section reads the U.S. Census Bureau, Vintage 2025 population estimates (NST-EST2025-ALLDATA, released January 2026, for July 1, 2024 to July 1, 2025): each state’s population and its natural change, net international migration and net domestic migration, with the rates per 1,000 residents as the Bureau states them. They are model estimates, not counts, and each vintage revises them. They sit beside the survey and are never mixed into it, because the two measure different things: the survey asks where people lived a year earlier, in calendar 2024, with sampling error; the estimate models July 2024 to July 2025 from administrative records. Where the two point in different directions, the page says so, and neither overrides the other.
The share of arrivals from bordering states divides the flows from states sharing a land border with the state by all flows into it in the table, using point estimates; each flow carries its own margin. To cite a state page: YardPermit, “<State> interstate migration, ACS 2024”, the page’s address — derived from U.S. Census Bureau ACS 2024 1-year Table T13.
What is published from the table
51 state hubs carry each state’s totals, national ranks and net balance. Of the state pairs, those with an estimate of at least 5,000, a margin no larger than 50% of the estimate, a reverse flow and a computable centroid distance, 368 pairs in this vintage, are ranked on the largest state-to-state moves table. The publication gate page reports how many pairs each condition excludes.
What is never published from it
County, metro, city and ZIP figures, because the table is state-level; driving distances, because centroid distances are straight lines between state centroids computed from TIGER geometry and say nothing about roads; prices, fuel and tolls, because they are not in the file; and any 2025 or later figure until the Census Bureau publishes it. State centroids are computed with d3.geoCentroid on TIGER-derived polygons. Distances shown on state-to-state pages are state-centroid straight-line distances. They are not driving miles, not a highway recommendation, and not a shipping zone. Driving distance requires a city pair and a routing provider.