Yes — active polling and aggregation for the 2026 U.S. midterms is well underway. Senate, House, governor, and generic congressional ballot surveys are being released weekly. House district polling remains much thinner than Senate or governor polling; only competitive districts tend to get repeated surveys, which is why the national generic ballot matters for estimating House control.
Below we classify sources into three types: pollsters (organizations that field surveys), aggregators (sites that collect and average others' polls), and forecast models (sites that blend polls with fundamentals, expert ratings, and sometimes prediction markets). Bias ratings describe editorial or historical lean, not whether a single poll is "correct."
Poll aggregators and polling averages
These sites do not usually conduct their own voter interviews. They compile public polls and compute averages or trend lines.
RealClearPolling
Type: Aggregator · Bias: Lean right · URL: realclearpolling.com
Probably the most useful single starting point for raw 2026 polls. Lists individual surveys by race — Senate, governors, generic congressional ballot — often within days of release. August 2026 feeds have included Texas Senate (Paxton vs. Talarico), North Carolina Senate, Michigan Senate primaries, Arizona governor, Wisconsin governor, and many others. Sister site RealClearPolitics is editorially conservative; the polling vertical is an aggregator, but headline selection on the broader RCP network can feel right-leaning.
Race to the WH
Type: Aggregator + forecast model · Bias: Neutral · URL: racetothewh.com
Strong choice if you want averages, not just individual polls. Separate 2026 sections for Senate, House, governors, and the generic ballot. Weights polls partly by recency, sample size, and pollster track record. Also runs simulation-based forecasts for chamber control. Independent project; not tied to a cable network.
PollingSource
Type: Aggregator · Bias: Neutral · URL: pollingsource.com
Newer tracker that monitors Senate, House, governors, primaries, and the generic ballot from public sources. Uses a 30-day polling average for race ratings when enough surveys exist; falls back to expert consensus (Cook, Sabato, Inside Elections) when data is thin. Useful map-based overview of projected seat counts.
Silver Bulletin (FLIPR averages)
Type: Aggregator + forecast model · Bias: Neutral · URL: natesilver.net
Nate Silver's FLIPR model combines polling averages with fundamentals and expert ratings. Polling averages adjust for registered vs. likely voters, pollster house effects, and poll age. Draws many race-level polls from the New York Times poll database. Analytical, probabilistic framing rather than partisan cheerleading.
270toWin
Type: Aggregator / map tool · Bias: Neutral · URL: 270towin.com
Best known for presidential maps; also hosts Senate power rankings, prediction-market snapshots, and links to external forecasts (e.g., Fox News Senate rankings). Does not field polls; aggregates ratings and market prices for interactive maps.
USPollingData.com
Type: Aggregator · Bias: Neutral · URL: uspollingdata.com
Battleground-focused tracker with Senate averages and news-style analysis. Smaller than RCP or Race to the WH but useful for quick state-level tables. Treat as supplemental, not definitive.
Forecast models (polls + fundamentals + markets)
These go beyond averaging polls to estimate win probabilities and seat counts.
Decision Desk HQ (DDHQ)
Type: Forecast model + results provider · Bias: Neutral · URL: votes.decisiondeskhq.com
Launched its 2026 congressional forecast in mid-July 2026, blending structural fundamentals, public polling averages, and prediction-market signals (Kalshi, Polymarket). Simulates House and Senate outcomes; also known for election-night race calls. As of launch, model showed divided government as plausible — Democrats favored for House, Republicans narrow favorites for Senate — but probabilities shift with new data.
FiftyPlusOne
Type: Forecast model · Bias: Neutral · URL: fiftyplusone.news
Bayesian ensemble combining polls, historical fundamentals, and expert race ratings (Cook, Sabato, etc.). Simulates tens of thousands of elections jointly for House and Senate. August 2026 runs showed Democrats strongly favored for House control with Senate closer to a toss-up — illustrative of how models can diverge from raw poll averages.
Expert rating shops (not pollsters)
These do not release voter surveys but rate races; aggregators often cite them when polls are sparse:
- Cook Political Report with Amy Walter — Nonpartisan analyst · Neutral. cookpolitical.com
- Inside Elections — Nonpartisan analyst · Neutral. insideelections.org
- Sabato's Crystal Ball (UVA Center for Politics) — Academic / analyst · Neutral. centerforpolitics.org
- Ballotpedia — Encyclopedia; compiles others' ratings · Neutral. ballotpedia.org
Pollsters: organizations that field surveys
These conduct interviews (phone, online, or mixed) and release results — often sponsored by media outlets.
University and nonprofit pollsters
- Quinnipiac University — Pollster · Neutral. Frequent state polls in the Northeast and Florida.
- Marist College (NPR/PBS partner) — Pollster · Neutral. National and New York–area surveys.
- Monmouth University — Pollster · Neutral. Detailed crosstabs; strong in New Jersey.
- Emerson College Polling — Pollster · Neutral to slight Republican house effect in some cycles. Active in 2026 primaries and battlegrounds (e.g., Iowa).
- Suffolk University — Pollster · Neutral. State polls including Iowa 2026 Senate/governor surveys.
- Siena College (NYT/PPH partner) — Pollster · Neutral. Multi-state battleground batteries with the New York Times.
- Pew Research Center — Pollster (nonpartisan research) · Neutral. Attitudes and turnout, less horse-race focus.
Media-sponsored national pollsters
- Fox News / Beacon Research & Shaw & Company — Pollster (bipartisan field team) · Poll neutral; sponsor lean right. August 2026 battleground polls covered Georgia (Ossoff–Collins), Maine (Collins–Jackson), Michigan (Rogers–El-Sayed), Ohio, and Texas.
- ABC News / Washington Post (Langer Research) — Pollster · Neutral methodology; center-left media sponsor.
- NBC News / Wall Street Journal (Hart Research + Public Opinion Strategies) — Pollster · Neutral. Bipartisan questionnaire team.
- CNN / SSRS — Pollster · Neutral. Battleground batteries and national tracking.
- Economist / YouGov — Pollster · Neutral. Weekly tracking; large online panel.
- Reuters / Ipsos — Pollster · Neutral. High-frequency online tracking.
Commercial pollsters (watch sponsor and house effects)
- Gallup — Pollster · Neutral. Approval and mood; limited district horse-race work.
- Rasmussen Reports — Pollster · Lean right. Historically pro-Republican house effects; still cited on approval tracking.
- TIPP / IBD — Pollster · Neutral to lean right on some national indicators.
- Fabrizio/Anzalone (Republican + Democratic firms paired) — Pollster · Neutral when bipartisan; individual releases may be campaign-adjacent.
State-based pollsters (examples active in 2026)
- UT Austin Texas Politics Project / YouGov — Pollster · Neutral. Texas statewide races (e.g., Talarico vs. Paxton).
- Mitchell Research — Pollster · Neutral. Michigan-focused.
- Hendrix College / Talk Business — Pollster · Neutral. Arkansas statewide surveys.
Practical tips for August 2026
- For raw polls by race: Start with RealClearPolling, then check Race to the WH averages.
- For chamber-control probabilities: Compare DDHQ, Silver Bulletin FLIPR, and FiftyPlusOne — they will disagree; that disagreement is informative.
- For House control: Lean on the generic congressional ballot plus district fundamentals; individual House polls exist mainly in competitive seats.
- For bias checks: Note who sponsored the poll, whether the firm is partisan, and how the result compares to other pollsters in the same race the same week.
If you are building a spreadsheet of 2026 races, pairing RealClearPolling (individual polls) with Race to the WH or PollingSource (averages) beside each Senate, competitive House, and governor row is a solid workflow. Add a forecast column from DDHQ or FLIPR if you want probability estimates — not just margin snapshots.