See where people moved to and from each state in one year.
Try the project ↓Help
The Census Bureau asks where people lived one year earlier. The answers help estimate how many people moved between states, but they are not exact counts.
Compare its estimated moves in and out, then pick a partner state from the two-way bars.
Read the ± margin beside each estimate before deciding that one direction was larger.
Pick a point to see which states had similar totals and which sat far from the diagonal.
Switch between nationwide leaders and the strongest pairs involving your selected state.
See where people moved to and from each state in one year.
The Census Bureau asked where people lived a year earlier. These are survey estimates for moves among the 50 states and Washington, D.C., not exact counts.
Choose a state. Each row shows estimated moves in both directions. Longer bars mean more people, and the printed ± value is the 90% margin of error.
The plausible ranges are 125,109–142,793 moving in and 122,210–139,246 moving out. Overlap means the survey does not support a simple claim that one total was larger.
Read each line from left to right. The total below combines both estimates; it does not count people who moved within a state.
What to notice: A strong connection can run in both directions. The larger bar is not automatically a clear difference when the margins of error overlap.
Each labeled point is one state or Washington, D.C. Right means more people estimated to have moved out; up means more estimated to have moved in. Points near the diagonal had similar totals.
What to notice: California sits far to the right because it had the largest estimated outflow, while Florida sits highest because it had the largest estimated inflow. Size alone does not explain why people moved.
The ranked bars add the two published directions for each state pair. The solid half runs from the first state to the second; the striped half runs back.
What to notice: California and Texas form the largest displayed pair at 122,608 estimated moves. A busy pair can still lean strongly in one direction.
This survey shows where people lived at two points one year apart. It does not say why they moved, follow the same people over time, or count moves within a state. Puerto Rico, moves from abroad, and 255 state-to-state directions with too few sample cases are excluded from these comparisons.
Why this exists
Which state pairs had the largest estimated two-way movement in 2024, how did each direction differ, and where did a chosen state's total moves in and out sit among the 50 states and Washington, D.C.?
Start with Indiana, choose any state, select one of its strongest state pairs, compare all states by estimated moves in and out, and switch the corridor ranking between nationwide leaders and pairs involving the selected state.
Method and limits
I selected the U.S. Census Bureau's 2024 State-to-State Migration Flows workbook through the allowlisted Census provider. The official release says the statistics come from the 2024 American Community Survey and Puerto Rico Community Survey one-year estimates. I opened the underlying workbook and its notes rather than relying on the catalog page. Census Bureau data and works created by agency employees generally are not subject to U.S. copyright protection under the agency's public-access policy; the page carries Census attribution and links the official policy.
I retrieved `State_to_State_Migration_Table_2024_T13.xlsx` on October 9, 2026. The 190,028-byte workbook has SHA-256 `9fcbca6df66289bd13fa7b95ac07aaeaeb7adac448458ff09273f37f014c4c85`. Its main table contains one estimate and one 90% margin of error for each published direction among the 50 states, Washington, D.C., Puerto Rico, U.S. Island Areas, and foreign countries. The two supplemental sheets provide authoritative total interstate moves into and out of each state.
The checked-in standard-library Python reducer verifies the workbook byte count, SHA-256, sheet lengths, 2,550 directed state combinations, and all 51 supplemental state totals. It excludes Puerto Rico, moves from abroad, U.S. Island Areas, and within-state moves so every comparison uses the same system of 50 states plus Washington, D.C. The transformation stores 51 state totals and 1,275 unordered state pairs. The 255 directions marked `N` by the Census Bureau are retained as unavailable rather than changed to zero. The published ranking uses the 1,048 pairs with estimates available in both directions. Repeated reductions produced the same 38,808-byte snapshot with SHA-256 `f23ecc05b37b6f932d7cd06760f1e7b1411cc73a0872ddabb74270f6a16e84a3`. The page imports that snapshot and makes no runtime request.
State totals use the supplemental estimates and margins of error exactly as published. Each two-way total adds the two directional point estimates. It is a descriptive sum, not an estimate with a newly calculated confidence interval. The interface prints each direction's published margin separately and avoids declaring a difference clear when ranges overlap. Bar length and scatter position encode estimates; direction labels, patterns, state abbreviations, and printed values keep the core comparisons readable without color.
These data show a person's state at two points one year apart. They do not reveal why someone moved, every move made during the year, moves within a state, or a continuing trend. Estimates come from a sample and also have nonsampling error that the displayed margins do not capture. State size, housing, jobs, schools, family ties, military assignments, disasters, and many other factors may shape the pattern; this project does not isolate any cause or judge whether a place gained or lost in a broader sense.