Election forecasts you can inspect.Tested models, shown with their sources and their uncertainty.

Updated October 22, 2026, 09:28 UTC

State legislatures › Colorado House

Colorado House District 21

2026 forecast · Updated October 22, 2026, 09:28 UTC

10% chance of a Democratic win · Brenda Miller (R), Michelle Tweed (D)

Colorado House District 21: ForecastColorado House 15: chance of a Democratic win 11%; 2024 president R+14Colorado House 17: chance of a Democratic win 98%; 2024 president D+11Colorado House 18: chance of a Democratic win 94%; 2024 president D+12Colorado House 47: chance of a Democratic win <1%; 2024 president R+31Colorado House 56: chance of a Democratic win <1%; 2024 president R+50Colorado House 60: chance of a Democratic win <1%; 2024 president R+36Colorado House District 21: Forecast60564721
Safe DLean DToss-upLean RSafe RHatched: not up in 2026 (shaded by holder) · dashed: other state

District 21 is likely Republican. Of its 6 neighbouring districts, 2 favour Democrats, 4 favour Republicans.

StatusForecast
Chance of a Democratic win10%
Likely margin (80% range)R+22 to even, middle R+11
CandidatesBrenda Miller (R), Michelle Tweed (D)
Candidate list fromofficial
Current memberMary Bradfield

How it has voted

On its current lines. Legislative contests and the top-of-ticket vote inside the district.

Legislature 2022R+17
Governor 2022even
Legislature 2024R+16
President 2024R+15
← RepublicanDemocratic →

Who lives here

Census Bureau, American Community Survey 2019-2023.

This districtColorado
Median household income$81,454$92,470
Bachelor's degree or more (25+)28%45%
Median age2938
White (not Hispanic)56%66%
Black10%4%
Hispanic24%22%
Asian3%3%
Born abroad6%10%
Jobs in manufacturing5%7%
Jobs in farming or mining1%2%

Initial version. This model was tested on every state legislative election from 1994 to 2024 before publication, forecasting each as if it had not happened. It beat the simple rule “the party that holds the seat keeps it”, and its chamber seat counts were off by 2.9 seats on average (target: under 3). Districts with no two-party race last time are read from their own presidential vote, a change tested in advance that cut their average error from 17 to 13 points. It missed one of its five accuracy targets: in 2024, the calmest year on record, its ranges were wider than they needed to be (92% of results fell inside its 80% ranges; target: at most 88%). Treat the ranges as cautious. The full test record and code.