Congress lives most years underwater with the public — approval in the teens, trust near record lows. The members are a different story. Across 27 straight elections, the vast majority of incumbents who asked voters for another term were handed one. Toggle the chart to see how little the public's verdict changes the result.
Averages across all 27 cycles, 1972–2024. Reelected is the share of incumbents who sought reelection and won; attrition is the complement — the share turned out in a primary or general election. Approval is Gallup's Congress job-approval reading nearest each election.
◗ The incumbency paradox
Voters detest the institution in the abstract, then re-hire the name they recognize.
That gap is the whole story — and the Both layer above draws it literally. Approval sinks toward the teens while reelection holds in the 90s. The money follows the incumbent, district lines are often drawn to protect them, and the challenger rarely gets a fair hearing, so a body the public rates in the teens keeps returning nearly all of its members.
Over these 27 cycles, House incumbents seeking reelection won an average of 93.8% of the time and Senate incumbents 83.8%, while Congress averaged just 28% approval at election time. In 2024 the split was at its starkest: 16% approved of Congress, yet 96% of House incumbents who ran were sent back.
| Year | House reelected | House attrition | Senate reelected | Senate attrition | Congress approval |
|---|
Reelection / attrition. Each figure is the share of incumbents who sought reelection and won. Members who retired, ran for another office, or died are excluded from the denominator; incumbents defeated in a primary or a general election count as losses. "Attrition" is 100% minus the reelection rate. 1972–2012 from Brookings/AEI, Vital Statistics on Congress, Tables 2-7 (House) and 2-8 (Senate); 2014–2024 compiled to the same definition from Ballotpedia, the UVA Center for Politics (Sabato's Crystal Ball), and OpenSecrets.
Approval. Gallup, "Do you approve or disapprove of the way Congress is handling its job?" — the reading nearest each November election. Gallup measures Congress as a whole and does not publish separate long-run House and Senate job-approval series, so the approval layer is a single line. The trend begins in 1974, and 1970s–80s readings are sparse (a few cycles use the only survey Gallup fielded that year); 1972 and 1984 have no comparable reading and are left blank.
On the 535. Only about 469 of the 535 seats are on any given ballot — all 435 House seats but just one-third of the Senate — so these rates describe the members actually facing voters each cycle, not the whole body at once.
Compiled for evangillissie.com · Data through the 2024 general election