U.S. Senate

Research Co., Nov. 2, 2026

Fielded October 31 to November 2, 2026 among 450 likely voters in Texas; online survey weighted to U.S. census figures for age and gender; margin of error ±4.6 points. First-choice result among all 450 respondents; the release's headline 51% to 48% figure counts decided voters only (433 decided). No sponsor is named in the release, which says only that the poll was conducted by Research Co. Display only: below the 500-respondent statewide minimum.

Emerson College, Nov. 1, 2026

Emerson College Polling for Poll Vault, the RealClearPolitics newsletter by Sean Trende (a media sponsor). Fielded October 31 to November 1, 2026; 500 likely voters. Talarico 49.8%, Paxton 48.4% as shown on the 99 Chambers pollster listing, which links the Poll Vault post (https://pollvault.substack.com/p/new-poll-vault-emerson-senate-and). The full release is for paid subscribers only, and it was not found in the public sources checked on November 3 (Emerson's site and Poll Vault's free page), so the margin of error, mode, question wording, ballot type, and undecided share are not available here. Display only: TexPolls averages require a public source and disclosed methodology.

New York Times / Siena College, Nov. 1, 2026

Fielded Oct. 27 to Nov. 1, 2026 among 1,215 likely voters. Live cellular and landline interviews in English and Spanish; 14% of interviews among self-reported Hispanics were in Spanish. Stratified sample of registered voters from the L2 voter file, supplemented with Marketing Systems Group matched cell numbers. NYT weighted with energy balancing to the likely electorate (party, race, age, gender, education, turnout history, region, 2024 vote) and adjusted for self-reported vote intention. Published full-sample margin of error ±3.3 points includes design effect 1.34; Siena's release gives ±3.4 (design effect 1.29) for the completed-interview sample, which applies to later questions, not these ballot results. Another candidate was reported as less than 0.5%, not an exact zero. First-choice result before the leaner follow-up: Paxton 41%, Talarico 48%, another candidate under 1%, don't know/refused 10%. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

AtlasIntel, Nov. 1, 2026

Fielded October 28 to November 1, 2026 among 1,678 Texas adults (not likely voters); web-based Atlas Random Digital Recruitment; margin of error about 2 points at 95%. First-choice result; 0.7% said they would not vote. No sponsor is named in the release. Display only: adult sample.

The Trafalgar Group, Nov. 1, 2026

Fielded October 31 to November 1, 2026 among 1,089 likely general-election voters; response rate 4.5%, 95% confidence. Ted Brown 2.4%. No sponsor is named. The report links only to Trafalgar's general methodology page and gives no survey-specific mode or weighting. Display only, the same treatment as Trafalgar's earlier Texas rows.

KStrat, Oct. 30, 2026

KStrat (Kreate Strategies, Houston) statewide survey, conducted October 27 to 30, 2026, among 900 likely Texas voters "using an online panel methodology"; published margin of error ±3.3%. Released October 30, 2026. The release reports 6% undecided in the Senate race and gives no figure for the Libertarian or any other candidate. It does not publish question wording, weighting, or the panel source; the full results and methodology are offered only after submitting a name and email on the release page, so they were not reviewed. Display only: held out of the average because the methodology is not publicly disclosed. The release names no sponsor or client.

YouGov, Oct. 29, 2026

YouGov release on YouGov's document host, fielded October 26 to 29, 2026 among 2,141 Texas likely voters; published margin of error ±2.9 points. Posted November 1, 2026 (the file is dated that day on YouGov's document host); the document names no sponsor, and no release article was found. Sample: "Stratified random sample of voters selected from a list of Texas registered voters who were contacted via text message and interviewed on YouGov's web-based survey system." Weighted by raking to gender, age, race, in-state geography (using the current Texas voter list), and imputed education, party score, and 2024 presidential vote; likely voters are defined by a model using past vote, demographics, and self-reported likelihood to vote. Question: "Who did you vote for, or who will you vote for, in the election for U.S. Senator?" Respondents who had already voted were asked how they voted; undecided voters were asked which candidate they lean toward, and the published result includes leaners. No initial ballot is published. Other combines someone else 1% and I would not vote 1%; not sure 1%. Counted in the average: unlike the September 28 to October 5 wave, which is held out, this release discloses its sample frame, mode, and weighting.

ActiVote, Oct. 28, 2026

Field October 22 to 28 (median October 25); opt-in app panel; two-way share summing to 100. Display only: below the 500-respondent statewide minimum.

Quantus Insights, Oct. 28, 2026

Fielded October 26 to 28, 2026 by SMS invitation to mobile phones drawn by stratified random sampling from active registered Texas voter-file records; raked on sex, age, race and ethnicity, education, metro type, and congressional district. 1,221 likely voters completed; 14 who said they would not vote in this race were removed, leaving 1,207 for the ballot. Published ±3.2 points for the full sample at 95 percent, design effect 1.28. Question: "If the election for U.S. Senate in Texas were held today, for whom would you vote?" The 27 undecided were asked "Even if you are not sure, which candidate for U.S. Senate do you lean toward?" Initial ballot Paxton 48.7, Talarico 47.6, Ted Brown (Libertarian) 1.4, undecided 2.2. With leaners Brown 1.4, undecided 0.8. Political sponsor classification: Quantus Insights sells campaign consulting to candidates and committees (firm homepage, Team section), and FEC records show the NRSC paid Quantus Insights (Andover, MN) $15,000 for "POLITICAL STRATEGY CONSULTING" on March 23, 2026. The release names no sponsor for this survey.

A2 Insights, Oct. 27, 2026

A2 Insights Texas Statewide Survey, fielded October 23 to 27, 2026, 631 likely voters. Online opt-in panel in English; screened to registered voters who already voted or will probably vote. Weighted with a statistical model on age, sex, race, education, party identification, and recalled 2024 vote to targets from Census data and exit polls. Published ±4.9 points at 95 percent using the effective sample size after the design effect. Question: "If the 2026 U.S. Senate election in Texas were held today and these were the candidates, who would you vote for?" Choices were the two nominees, other candidate, and undecided; no Libertarian named and no leaner follow-up. Other candidate 2.3, undecided 0.3. The report lists the sponsor as N/A; A2 Insights, LLC is an Ann Arbor, Michigan firm and AAPOR member. Released October 28, 2026 on A2's Substack.

CNN / SSRS, Oct. 27, 2026

CNN poll conducted by SSRS, Oct. 20 to 27, 2026 (released Oct. 29), online and by telephone among 923 Texas registered voters (L2 registration-based sample n=423 plus SSRS Opinion Panel n=500), in English or Spanish. Likely-voter results weight all registered voters by modeled probability of voting. The margin of sampling error was +/-4.2 points including design effects. The published figure combines the initial question with a lean follow-up and, for those who already voted, how they voted. No initial-only ballot is published. Neither 5%, other 3%, no opinion under 0.5%. Senate wording: "If the election for U.S. Senate were held today and the candidates were James Talarico, the Democrat, and Ken Paxton, the Republican, who would you be more likely to vote for?" followed by "If the election for U.S. Senate were being held today, as of today, do you lean more toward:". Counted in the average under the leaner rule adopted Oct. 21, 2026.

On Point Polling, Oct. 27, 2026

Held out of the average: the weighting target ("the On Point composition") isn't published, undecided voters are assigned by model, and the firm, The Public Sentiment Institute (doing business as On Point Polling), was permanently banned by the aggregator FiftyPlusOne in August 2026 over an earlier Florida poll. Released October 29, 2026 on On Point's Substack. 1,184 Texas registered voters interviewed online through a Cint opt-in panel, with quotas on age, gender, and location. The release gives October 27 as the field date; the crosstab title says October 27 to 28. Likely voters are the same 1,184 respondents reweighted by a turnout model, with no separate likely-voter count. Published ±2.9 points for likely and registered voters. Ballot wording not published. This row stores the result as asked; On Point's "undecideds leaned" result allocates undecided voters "based on how decided voters with similar views and backgrounds plan to vote," not by a leaner question. Likely voters as asked: Talarico 50.6, Paxton 44.1, Ted Brown (Libertarian) 2.2, undecided 3.2. Model-allocated: Talarico 51.7, Paxton 45.8, Brown 2.3, undecided 0.2. Registered voters as asked: Talarico 44.6, Paxton 34.7, undecided 17.7. The methodology tab states: "Sponsor: On Point Polling. Not paid for by any candidate or committee."

254 Labs / YouGov Blue, Oct. 26, 2026

Fielded October 21-26, 2026 by YouGov Blue 'on behalf of 254 Labs'; released October 28. 1,000 Texas registered voters interviewed online and by text-to-web, quota-sampled, then weighted to likely-voter targets that 254 Labs supplied (including an early-vote turnout benchmark); the release headlines the results as likely voters, but no likely-voter count is published. Margin of error about 4.6 points. Respondents who had already voted were asked how they voted; undecided respondents were asked which way they lean and the lean answers are included in the totals. The release publishes no question wording, no first-choice result, and full crosstabs are subscriber-only. Who paid for the poll is not disclosed; listed as political because 254 Labs describes itself as a Texas political research and analytics organization that serves 'mission-aligned subscribers,' not because a payer was shown. Shown for reference and left out of the headline average.

Data for Progress, Oct. 26, 2026

Data for Progress crosstabs: "From October 23 to 26, 2026, Data for Progress conducted a survey of 712 likely voters in Texas using SMS and web panel respondents." Weighted to likely voters by age, gender, education, race, geography, and recalled presidential vote; English only; "The margin of error associated with the sample size is ±4 percentage points." The release (October 30, 2026) says DFP surveyed Maine, Michigan, and Texas "from October 23-27"; the Texas crosstabs give October 23 to 26, used here. Question: "If the November 2026 general election for U.S. Senate were held tomorrow, and these were the candidates, who would you vote for?" Choices: Talarico, Paxton, Libertarian Ted Brown, not sure; no leaner follow-up published. Ted Brown (Libertarian) 2, not sure 4. Political sponsor classification: Data for Progress describes itself as "a progressive research organization" and offers campaign clients "campaign strategy"; FEC records show 2025-26 polling payments from candidate committees (Sage for Senate $15,000, Yassamin Ansari for Congress $8,000, Nate Blouin for Utah $6,000) and PACs including Justice Democrats PAC and Working Families Party PAC. The release names no client for this survey.

Texas Southern University, Oct. 25, 2026

Fielded October 21-25, 2026; released October 29. The Barbara Jordan Public Policy Research and Survey Center at TSU reports a probability-based sample of 2,000 likely voters, +/-2.19-point margin of error, and a two-part likely-voter screen supplemented by past election participation. A random sample of registered voters was contacted by SMS and directed to an online survey in English or Spanish; weighted by multidimensional calibration on demographic, historical electoral, and geographic benchmarks. A little more than one-third of respondents had already voted, and the report calls the result vote intention. The report names no outside sponsor, publishes no question wording, and describes no leaned follow-up. Libertarian Ted Brown 3%, unsure 1%.

N+ Univision / YouGov, Oct. 21, 2026

Fielded October 13-21, 2026; released October 27. N+ Univision media survey, overseen by Dr. Sergio Garcia-Rios (N+ Univision Director of Polling and Data) with BluePoint Analytix assisting on questionnaire review and analysis; fieldwork by YouGov. Online interviews of 1,000 registered Texas voters, weighted to registration lists, Census data, 2022 turnout, and 2020 and approximate 2024 presidential vote. Registered voters, so excluded from the likely-voter headline. The crosstabs report a 95% margin of error of 3.8 points computed with a design-effect adjustment for the weights, and note it does not reflect non-sampling error such as panel selection bias. Crosstabs label this measure 'Full Recoded' and publish neither ballot wording nor a separate leaner follow-up; one result only, treated as first choice. 1% chose another candidate and 5% were undecided; published shares total 101% from rounding.

ActiVote, Oct. 21, 2026

Field October 17 to 21 (median October 19); opt-in app panel; two-way share summing to 100. Display only: below the 500-respondent statewide minimum.

Rasmussen Reports, Oct. 21, 2026

Rasmussen Reports toplines: 965 Texas likely voters, October 20-21, 2026, ±3 points at 95% confidence; fieldwork by Pulse Opinion Research. Rasmussen says it pays for its own polls. Question: "Republican Ken Paxton, Democrat James Talarico, or Libertarian Ted Brown?" Brown 5%, not sure 5%; no leaner follow-up. The toplines page and the article agree on the sample, dates, and numbers, unlike the September release, which is still held over its conflicting sample and state. The article calls it a "telephone and online survey" and links only Rasmussen's firm-wide methodology page; it gives no mode mix or weighting for this survey, and the article's dateline reads September 30, 2026. Display only: held out of the average because survey-specific mode and weighting are undisclosed, the same treatment as the September Trafalgar and October YouGov polls.

VoteHub / Verasight, Oct. 20, 2026

Sample: 601 actual registered-voter responses (600 complete and one partial), modeled to a likely-voter electorate; effective LV n=508.5 (rounded 509), not 509 actual LV interviews. Fielded October 1 to 20, 2026; released October 23. Verasight collected the email-invited online survey; VoteHub weighted and analyzed it. The Verasight Community mixes address/text probability recruitment and targeted online recruitment; this is not a simple random probability sample. Reported margin of error ±4.3 points at 95% confidence; design effect 1.18. Exact ballot/leaner wording, who received the follow-up, detailed weighting targets/trimming, and the likely-voter model formula are not disclosed in the reviewed materials. VoteHub is the identified media partner and publisher; no political commissioner is named. Exact payer and funding amount are not disclosed. We list it as a nonpartisan poll based on who published it; we have not reviewed its funding. Published LV leaned result: Talarico 47.6%, Paxton 45.2%, Ted Brown 1.4%, unsure 5.4%, missing 0.4%. Our table has no column for missing responses, so the 0.4% is noted here instead of being added to unsure or other. Initial support: Talarico 45.6%, Paxton 41.4%, Brown 3.8%; the report doesn't say what the other 9.2% of first-choice answers were, so we don't count them as undecided. The first-choice and registered-voter results come from the same survey and are not counted as separate polls. Brown falls from 3.8% to 1.4%, so without the questionnaire the follow-up cannot be assumed to concern only initially undecided respondents.

Emerson College, Oct. 19, 2026

Fielded Oct. 17 to 19, 2026. Sponsored by Nexstar Media. Initial ballot Talarico 48, Paxton 46, Brown 1, undecided 5. Undecided respondents were asked which candidate they lean toward. MMS text-to-web, IVR landline, and PureSpectrum panel. The poll was in English and Spanish. Counted under the leaner rule adopted Oct. 21, 2026.

Fox News, Oct. 19, 2026

Fielded Oct. 15 to 19, 2026. The Fox News Poll was jointly directed by Beacon Research (D) and Shaw & Company Research (R). The 2,011 registered-voter sample was drawn from the statewide voter file. The likely-voter subsample is n=1,398 with a published ±2.5-point margin of sampling error. Interviews used live landline/cell calls and text-to-web, with English and Spanish available. Weights used age, race, education, and area. The ballot named Paxton and Talarico only. Respondents unsure after the initial question were asked which candidate they leaned toward, so published results include that follow-up. Respondents who had already voted early were asked how they voted. Other and would-not-vote responses were under one-half percent. Another 1% did not know. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

Texas Public Opinion Research, Oct. 18, 2026

Fielded October 15-18, 2026, via an online panel; n=969 likely voters, margin of error ±3.9 points; weighted by gender, race, age, education, income, geography, and 2024 vote recall. Shown with leaners (Q24, "Intermediate Push," after undecided respondents were asked which candidate they lean toward). First-choice result (Q22): Paxton 45%, Talarico 50%, undecided 3%. Libertarian Ted Brown 1% on the first choice and 2% with leaners; would not vote in that race shown as "-" in the topline. Still not sure after the lean question: 1%. TPOR's release headlines a "Final Push" result that asks remaining undecided respondents a second time and leaves none undecided; that forced result is not used. TPOR describes itself as a nonpartisan polling initiative and names no external commissioner. The site's copyright line names Luke Warford, whom the Texas Tribune (June 6, 2024) describes as the Texas Democratic Party's former chief strategy officer. This topline's page footer reads "Prepared by Slingshot Strategies"; September's read "Prepared by Texas Public Opinion Research."

Pollfinity Research, Oct. 18, 2026

Fielded October 15-18, 2026; released October 22. 278 registered Texas voters by text-to-web, drawn from a statewide voter file and weighted to the 2022 midterm electorate; margin of error ±6.5 points at 95% confidence (effective n=230). Pollfinity says it was not requested or paid for by any candidate, party, or organization; it is owned and self-funded by Lars Mapstead, a 2024 Libertarian presidential candidate. Libertarian Ted Brown 2%, someone else 2%, undecided 5%; the release reports one topline and no separate leaner follow-up. Registered-voter sample below the 500-respondent statewide minimum. Display only. Ballot wording not published.

CBS News / YouGov, Oct. 16, 2026

CBS News poll conducted by YouGov, Oct. 8 to 16, 2026, online among 1,492 Texas registered voters (YouGov panel plus 13 from other panels), in English or Spanish. The poll was weighted to Census benchmarks and 2024 presidential vote. The margin of error is ±3.4 points for registered voters and ±3.5 for likely voters (CBS News published the methodology with its article: https://www.cbsnews.com/news/texas-senate-talarico-paxton-poll/). Likely voters are modeled from stated intention, demographics, and vote history rather than a cutoff screen, so the release gives no unweighted likely-voter count. The 1,343 figure in some trackers does not appear in the release. Q6 (Senate) asks "If the election were held today, would you vote for:" among likely voters, including those leaning toward a candidate. Someone else and not sure are both published as 0% (rounded). Crosstab weighted bases (Senate 1,458, Governor 1,460) are weighted counts, not respondents. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

Cor Strategies / Heritage Action, Oct. 15, 2026

Fielded October 10-15, 2026 among 1,409 likely Texas voters; margin of sampling error +/-2.61 points. Mixed mode: live calls, online panels, and automated telephone interviews. Directed by Stefani Buhajla in partnership with Cor Strategies, LLC; the crosstabs state "Heritage Action paid for all costs associated with this survey." Display only: advocacy-group sponsor. The crosstabs are titled "TX Election Integrity," and the Senate question (Q11) came after 10 proof-of-citizenship questions, so question order may have affected answers. Q11 wording: "If the election for U.S. Senate were held today, for whom would you vote? The Democrat, James Talarico, or the Republican Ken Paxton?" Undecided 16%; no leaner follow-up shown. Weighted to a party-affiliation target of 42% Republican, 34% Democrat, and 24% Independent.

University of Houston Hobby School of Public Affairs, Oct. 14, 2026

Fielded Oct. 10 to 14, 2026 among 2,000 Texas likely voters contacted by SMS text and directed to an online survey, in English or Spanish. It was weighted by multidimensional calibration on demographic, historical electoral, and geographic benchmarks. The margin of error is ±2.19 points. The report and its Oct. 19 release publish no question wording and no separate questionnaire or topline document (as of Oct. 19, 2026, the Hobby School's Texas 2026 page links only this report, which prints the topline results). The report prints an undecided share and describes no leaner follow-up, the same disclosure standard as other polls in the average. No outside sponsor is named. Released Oct. 19, 2026. Libertarian Ted Brown 3%, undecided 3%.

InsiderAdvantage, Oct. 14, 2026

InsiderAdvantage release by Matt Towery posted Oct. 15, 2026, covering Texas, Kansas, and Maine. The Texas topline tab, posted to the @InsiderPolling X account on Oct. 16 (https://x.com/InsiderPolling/status/2111096358233129297), gives field dates Oct. 13 to 14, 1,200 likely voters, and a 2.83 margin of error. The release describes each survey as mixed-mode text and panel, weighted for age, race, gender, and political affiliation. No sponsor is named. The October question folds leaners into the ballot ('for whom would you vote, or toward whom are you leaning?'), unlike the September wave, which asked an initial ballot naming the three candidates with no lean prompt (5.1% undecided then, 1.0% now). Libertarian Ted Brown 2.4%. Published to one decimal. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

ActiVote, Oct. 14, 2026

ActiVote poll published October 14, 2026, field period September 30 to October 14 (median October 7), 400 likely midterm voters. Opt-in app panel: ActiVote users pick a preferred candidate on the race screen in its free voter-education app (English and Spanish), and the latest preference in the field window is counted. Weighted by demographics, voting chance from the L2 voter file, region, and ActiVote's own left-to-right 'Political Spectrum' score. ActiVote Inc., a Delaware public benefit corporation that describes itself as non-partisan, runs and funds every poll; no outside sponsor. Published as a two-way share that sums to 100 (Talarico 51.9, Paxton 48.1): no undecided or other-candidate figure is reported, and Libertarian Ted Brown does not appear. The page header prints 'MoE ±4.9%', but ActiVote's AAPOR disclosure describes that figure as an 'average expected error' from its past accuracy that 'should not be mistaken for the Margin-of-Error of a probabilistic poll', so we show no margin of error. Display only: below the 500-respondent statewide minimum, and the two-way share is not comparable with initial ballots that report undecided voters.

GS Strategy Group / New River Strategies (Cook Political Report), Oct. 13, 2026

Fielded October 9-13, 2026, in partnership with The Cook Political Report. Libertarian Ted Brown 3%, don't know 4%. Display only: no margin of error, mode, sample source, or weighting published.

Marist Poll, Oct. 11, 2026

Fielded October 8-11, 2026. Marist interviewed 1,196 Texas registered voters by live phone (265), text (698), and online (233), in English or Spanish; phone sample from L2's voter file, online respondents Cint-matched to it. Weighted by age, gender, income, race/ethnicity, education, and region. Likely-voter results apply a probability-of-voting weight to all registered voters, so Marist publishes no separate likely-voter count; the ±4.9-point likely-voter margin includes the design effect. Another party's candidate 2%, undecided 4%, volunteered no response 2%. Raw support, no leaner follow-up. Registered voters: Talarico 48%, Paxton 44%. The Marist Poll is an educational program funded by Marist University; no outside sponsor is named.

Quantus Insights, Oct. 11, 2026

Fielded October 9 to 11, 2026 by SMS invitation to mobile phones drawn by stratified random sampling from active registered Texas voter-file records; raked on sex, age, race and ethnicity, education, metro type, and media market. 854 likely voters responded; 15 who said they would not vote were removed, leaving 839 for the ballot. Published ±3.6 points for the full sample, design effect 1.18. Question: "If the election for U.S. Senate in Texas were held today, who would you vote for?" Ted Brown (Libertarian) 2.2%, undecided 3.9%; no leaner follow-up. Political sponsor classification: Quantus Insights sells campaign consulting to candidates and committees (firm homepage, Team section), and FEC records show the NRSC paid Quantus Insights (Andover, MN) $15,000 for "POLITICAL STRATEGY CONSULTING" on March 23, 2026. The release names no sponsor for this survey.

ReconMR / Siena Research Institute, Oct. 8, 2026

ReconMR / Siena Research Institute, fielded October 5 to 8, 2026 among 617 Texas likely voters (400 live phone in English and Spanish, 217 text-to-web); landline and cell sample from the L2 voter database, with quotas on region, race, and party. Weighted on party, age, sex, region, race/ethnicity, education, and stated and derived vote likelihood. Published ±4.4 points including the design effect. The TI disclosure names ReconMR as research sponsor, and the release describes ReconMR as "an independent, non-partisan research institute"; Siena's research institute co-presents the poll. Q3: someone else 2%, would not vote 1%, don't know/refused 5%. Initial ballot; no leaner follow-up is published.

The Trafalgar Group, Oct. 8, 2026

Fielded October 6 to 8, 2026 among 1,086 likely general-election voters; response rate 5.4%, ±2.9 points. Question: "If the election for U.S. Senate were tomorrow, for whom would you most likely vote?" Ted Brown (Libertarian) 1.8%, undecided 7.6%. No sponsor is named. The report links to Trafalgar's general methodology page, which lists six possible contact methods without saying which this survey used, and publishes no weighting. Display only: held out of the average pending survey-specific method disclosure, the same treatment as the September Trafalgar poll.

YouGov, Oct. 5, 2026

YouGov release on YouGov's document host, fielded September 28 to October 5, 2026 among 3,625 Texas likely voters; published margin of error ±3 points. The document names no sponsor and does not disclose mode, sample frame, or weighting. The UT Texas Politics Project tracker (texaspolitics.utexas.edu/blog/texas-2026-u-s-senate-poll-tracker) lists it as an in-house YouGov poll; a Wikipedia listing attributes it via The New York Times. The Senate question says undecided voters were asked their preference and allocated, so this topline includes leaners; no initial ballot is published. Other combines someone else 2% and will not vote in this race 1%. Display only: held out of the average because the survey method is undisclosed.

UT Texas Politics Project, Oct. 5, 2026

University of Texas / Texas Politics Project in-house poll (no outside sponsor), fielded Sept. 25 to Oct. 5, 2026. YouGov interviewed 1,253 Texas registered voters online and matched them to 1,200, weighted by propensity score and post-stratified on 2024 presidential vote, gender, age, race, and education. Likely voters (weighted base 861, 905 unweighted respondents) report the published ±3.34-point margin. Ted Brown (Libertarian) 1%, someone else 2%, not sure 0%. UT's published ballot folds in a forced follow-up ("If you had to make a choice, who would you choose?") among respondents who first said they didn't know, and the topline does not print the initial ballot. The initial ballot, computed from UT's posted microdata with its own likely-voter flag and weight, is Paxton 45.35%, Talarico 48.74%, not sure 3.72%. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

New York Times / Siena College, Sept. 30, 2026

Fielded Sept. 21 to 30, 2026 among 615 likely voters. Live cellular and landline interviews in English and Spanish. L2 voter-file sample supplemented with Marketing Systems Group matched telephone numbers. NYT weighted with energy balancing, demographic and turnout-history targets, and modeled/self-reported vote likelihood. Published full-sample margin of error ±4.5 points includes design effect 1.28. The smaller completed-interview sample (465, ±5.2 points) applies to later questions, not these ballot results. Another candidate was reported as less than 0.5%, not an exact zero. First-choice result before the leaner follow-up: Paxton 43%, Talarico 49%. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

Fox News, Sept. 28, 2026

Fielded Sept. 24 to 28, 2026. The Fox News Poll was jointly directed by Beacon Research (D) and Shaw & Company Research (R). The 1,203 registered-voter sample was drawn from the statewide voter file. The likely-voter subsample is n=881 with a published ±3.0-point margin of sampling error. Interviews used live landline/cell calls and text-to-web, with English and Spanish available. Weights used age, race, education, and area. The ballot named Paxton and Talarico only. Respondents unsure after the initial question were asked which candidate they leaned toward, so published results include that follow-up. No other candidate or would-not-vote responses were recorded. Another 1% did not know. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

UMass Lowell / YouGov, Sept. 28, 2026

UMass Lowell Center for Public Opinion survey fielded by YouGov, September 18 to 28, 2026, released October 8: 850 Texas likely voters, self-administered online, non-probability sample matched to a sampling frame and weighted to CPS voting supplement weights and 2024 presidential vote, with weights adjusted toward likely voters. Adjusted margin ±4.5 points. "The University of Massachusetts Lowell sponsored the research." Ted Brown (Libertarian) 2%, another candidate 1%, would not vote 1%, undecided 7%. Initial ballot; no leaner follow-up.

Guidant Polling & Strategy / Public First Action, Sept. 28, 2026

Guidant Polling & Strategy memo dated October 6, 2026: 800 likely general-election voters, September 20 to 28, 2026, mostly cell phone with text-to-web and landline interviews, ±3.46 points; weighting, sample frame, and likely-voter definition are not disclosed. The memo was released as images in a Semafor reporter's post (October 7) introduced as "New polls commissioned by Public First Action"; no copy appears on Guidant's or the sponsor's site. Public First Action is a 501(c)(4) whose founders announced two super PACs. Ted Brown (Libertarian) 4%, undecided 5%. The memo text gives Talarico 46.6% and Paxton 44.2%; the chart rounds to 47 and 44. Advocacy-group-sponsored poll, shown with the sponsor disclosed and excluded from the headline average.

Big Data Poll, Sept. 26, 2026

Big Data Poll for its Public Polling Project (crowdfunded, supplemented by Big Data Poll if necessary), released Sept. 27, fielded Sept. 24 to 26, 2026 among 698 likely voters (793 registered). The margin of error is ±4.0 points at 95% confidence. About 60% live phone, P2P SMS, and text-to-online from the L2 voter file and about 40% online via Lucid (Cint). It was weighted by sex, age, race and ethnicity, education, and geography. The release says the project is not funded by or affiliated with any candidate, campaign, or committee. Published to one decimal. With undecided voters asked to lean: Talarico 46.8%, Paxton 44.9%. Registered voters, initial: Talarico 43.6%, Paxton 39.3%. The release text gives no initial-ballot undecided or other shares. They are not recorded here. First-choice result before the leaner follow-up: Paxton 41.6%, Talarico 44.6%. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

Stratus Intelligence, Sept. 24, 2026

Stratus Intelligence memo, fielded September 22 to 24, 2026 among 865 likely voters; 81% SMS-to-web and 19% live phone; ±3.5 points. The memo names no sponsor and publishes neither question wording nor weighting. Breitbart (September 29) and the Wikipedia Senate-race listing report the poll was conducted for Texans for a Conservative Majority. Sponsored poll, shown with the sponsor disclosed and excluded from the headline average. Undecided 4%.

Texas Public Opinion Research, Sept. 22, 2026

Fielded September 19-22, 2026, via an online panel; n=1,007 likely voters, margin of error ±3.6 points. Shown with leaners (Q18). First-choice result (Q16): Paxton 42%, Talarico 48%. Libertarian Ted Brown 2%; would not vote in that race 1%. TPOR describes itself as a nonpartisan polling initiative and names no external commissioner. The site’s copyright line names Luke Warford, whom the Texas Tribune (June 6, 2024) describes as the Texas Democratic Party’s former chief strategy officer. Weighting used gender, age, race, education, 2024 vote recall, and geography.

Marist Poll, Sept. 20, 2026

Fielded September 17-20, 2026 among Texas registered voters age 18+. The published result is raw support, with no leaner follow-up. The sample used live phone (203), text (692), and online (244) interviews; English and Spanish; phone respondents were sampled from L2's voter file and online panel respondents were Cint-matched to that file. Results were weighted by age, gender, income, race/ethnicity, education, and region. The published ±4.4-point margin includes a 2.33 design effect. The Marist Poll is an educational public-opinion program funded by Marist University; no external commissioner is named.

Texas Southern University, Sept. 19, 2026

Fielded September 15-19, 2026. The Barbara Jordan Public Policy Research and Survey Center at TSU reports a probability-based sample of 1,800 likely voters, ±2.31-point margin of error, and a two-part likely-voter screen supplemented by past election participation. A random sample of registered voters was contacted by SMS and directed to an online survey in English or Spanish. The report does not name the sampling or survey vendor or any outside sponsor. The published table includes an explicit 4% unsure category and names Libertarian Ted Brown at 3%; it describes no leaned follow-up.

The Trafalgar Group, Sept. 17, 2026

The Trafalgar Group, fielded September 15 to 17, 2026 among 1,079 likely general-election voters; ±2.9 points at 95% confidence; 5.1% response rate. Results published to one decimal; Ted Brown 2.5%, undecided 5.9%. The report names no sponsor. It links to a firm-wide methodology page listing six possible modes (live calls, automated voice, text, email, digital dial-back, and online opt-in) without saying which were used for this survey or how it was weighted. Display only: held out of the average pending survey-specific method disclosure.

Emerson College, Sept. 14, 2026

Fielded September 12-14, 2026. Sponsored by Nexstar Media. Libertarian Ted Brown received 2%; 4% were undecided. The reported ±3.0 figure is a 95% credibility interval, not a traditional sampling margin of error. MMS text-to-web from a probability-based voter panel plus PureSpectrum; weighted by gender, education, race, age, and region; English only.

ReconMR / Siena Research Institute, Sept. 11, 2026

Fielded September 8-11, 2026. ReconMR/Siena's nonpartisan Texas Poll. Prompt named only Paxton and Talarico; 3% chose someone else, 1% would not vote, and 4% did not know or refused. Mixed live telephone and text-to-web voter-file sample; telephone interviews in English and Spanish; weighted to voter-file and demographic targets. MOE includes design effects.

NBC 5 Dallas / Telemundo Texas / Mason-Dixon, Sept. 10, 2026

Mason-Dixon Polling & Research poll for the Telemundo stations in Texas; NBC 5 Dallas publishes it as the NBC 5 Dallas/Telemundo Station Group Texas Voter Poll. Released September 21, 2026. Fielded September 8 to 10, 2026: 625 registered voters who said they were likely to vote in November, interviewed live by telephone after random selection from a telephone-matched voter registration list with landline and cell numbers; quotas were set to reflect voter turnout by county. The results document describes no weighting. Margin of error ±4 points at 95% confidence. Order-rotated ballot naming Libertarian Ted Brown (3%); 8% undecided. A separate 401-person Hispanic likely-voter survey (September 8 to 12, ±5) is not included in this row.

InsiderAdvantage, Sept. 9, 2026

InsiderAdvantage release by Matt Towery posted September 10, 2026, covering Texas, South Carolina, and Ohio. The Texas topline tab (posted separately) gives field dates September 8 to 9, 1,200 likely voters, and a ±2.83 margin of error. The release describes each survey as mixed-mode panel and text, weighted for age, race, gender, and political affiliation. No sponsor is named. Party-labeled ballot naming Libertarian Ted Brown (2.2%); 5.1% undecided. Published to one decimal. The release does not give the sample source or the share of interviews by mode. The ballot figures and field dates are on InsiderAdvantage's Texas topline tab: https://insideradvantage.com/top-line-tabs-texas/.

N+ Univision / YouGov, Sept. 4, 2026

Fielded August 27-September 4, 2026. N+ Univision media survey conducted online by YouGov. Registered voters, so excluded from the likely-voter headline. The nonprobability survey reports an equivalent margin of error for a matched propensity-score design, not a traditional sampling MOE. Crosstabs label this measure 'Combined - Recoded' and do not publish exact ballot wording or leaner treatment; 1% chose another candidate and 7% were undecided.

AARP / Fabrizio Ward + Impact Research, Sept. 1, 2026

Fielded Aug. 30 to Sept. 1, 2026. Commissioned by AARP and conducted by the bipartisan team of Fabrizio Ward and Impact Research. Published total includes a forced follow-up among initially undecided voters. The annotated questionnaire does not publish a clean, exact unallocated topline. In the two-candidate prompt, 8% remained undecided after the follow-up. Live cell 40%, live landline 15%, and SMS-to-web 45%. The poll was in English and Spanish. The sample came from a voter list. Sampling error is unadjusted for design effects. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

Overton Insights / Texas Public Policy Foundation, Aug. 26, 2026

Joint survey released Sept. 2, 2026 by the Texas Public Policy Foundation, which describes itself as a non-profit free-market research institute, and Overton Insights, which describes itself as a nonpartisan public opinion research firm. Text-to-web survey of 1,167 likely voters, Aug. 24 to 26, 2026. The margin of error is ±2.9 points. The toplines say the data were weighted using iterative weighting but do not list the weighting targets, the sample source, or the likely-voter definition. Initial two-way ballot, published to one decimal. The undecided share was 12.6%. After a forced lean question, the published cumulative result is 50.0% to 50.0%. Counts toward the average: the site’s written rule excludes campaign- and political-operative-sponsored polls, and a non-profit research institute is neither. First-choice result before the leaner follow-up: Paxton 43.4%, Talarico 44%. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

Texas Public Opinion Research, Aug. 24, 2026

TPOR director Luke Warford is a Democratic strategist. Ballot named Libertarian Ted Brown at 1%.

UT Texas Politics Project, Aug. 13, 2026

Ballot named Libertarian Ted Brown at 3%. Registered voters, not likely voters.

Emerson College, Aug. 10, 2026

Emerson College Polling/Nexstar Media, fielded Aug. 9 to 10. 5% undecided, 2% someone else.

Texas Southern University, July 30, 2026

Three-way ballot incl. Ted Brown (L) 2%, with 6% undecided. Barbara Jordan Public Policy Research and Survey Center. The poll used SMS-to-web, was fielded July 27 to 30, and was coauthored by Rice's Mark P. Jones. Sampling frame and weighting variables are not disclosed in the report.

Texas Pulse / ReconMR, July 30, 2026

Texas A&M Bush School consortium, fielded July 27 to 30, released Aug. 7. MOE is the core-question figure (n=619). The ballot is two-way. Talarico led 53-30 among independents.

Fox News, July 27, 2026

Beacon Research (D) and Shaw & Company Research (R), fielded July 23 to 27. Live interviewers on landlines and cellphones plus text-to-web, sampled from the statewide voter file. Registered voters, not likely voters. Both ballots are inside the margin of error.

Texas Public Opinion Research, July 17, 2026

TPOR director Luke Warford is a Democratic strategist; conducted with Lake Research Partners. 14% undecided.

New York Times / Siena College, June 27, 2026

Independent newsroom-sponsored poll. Fielded June 19 to 27, published June 30. The sample is n=656. Initial Senate response: Talarico 45%, Paxton 44%, another candidate 1%, undecided/refused 9%. Published rounded categories sum to 99%. The 47–47 result includes leaners, per the historical comparison in the subscriber-supplied September NYT toplines. First-choice result before the leaner follow-up: Paxton 44%, Talarico 45%. Counted in the average under the leaner rule adopted Oct. 21, 2026: when a poll publishes a result that includes leaners, the table and average use it.

UT Texas Politics Project, June 22, 2026

First nonpartisan poll after the runoff (fielded June 5 to 12 by YouGov). Registered voters. MOE is UT TPP's weight-adjusted figure (±2.83% raw). Also on the ballot: Ted Brown (L) 3%, undecided 10%. The race was within the margin of error.

Texas Public Opinion Research, May 27, 2026

Commissioned by Luke Warford (2022 Democratic nominee for Texas Railroad Commissioner) via Lake Research Partners.

Texas Public Opinion Research, April 20, 2026

TPOR director Luke Warford is a Democratic strategist; commissioner not disclosed.

UT Texas Politics Project, April 20, 2026

Large undecided / no-opinion share.

Emerson College, Jan. 12, 2026

Emerson College Polling/Nexstar Media, fielded January 10-12, 2026. Sponsored by Nexstar Media. 9% undecided. The reported ±2.8 figure is a credibility interval, not a traditional sampling margin of error.