Early voting for the November 3, 2026, midterm elections is underway in most U.S. states, and participation in the first days has run well above previous midterm cycles. That combination of ballots already being cast plus unusually high numbers is what has pushed the subject into wide discussion. According to the NBC News early-vote tracker, about 7,764,793 mail-in and early in-person ballots had been cast nationally as of its October 17, 2026, update, split into roughly 5.7 million mail-in and 2.1 million in person. The same tracker reported 54,848,197 mail-in ballots requested or sent as of that date.
The NBC News tracker also reported the party registration mix of the ballots already cast as of October 17, 2026: about 53 percent from registered Democrats, 31 percent from registered Republicans, and 16 percent from other or unaffiliated voters. Among the 54,848,197 mail ballots requested or sent, the breakdown was roughly 48 percent Democratic, 30 percent Republican, and 22 percent other. These figures matter because they describe who has voted so far, not who will win. The dataset updates daily, so any single day's snapshot is provisional.
The New York Times early-voting tracker notes that more than half of all U.S. voters now vote before Election Day, a structural change from earlier decades. That shift means the weeks before the first Tuesday in November increasingly carry the weight that Election Day itself once carried. Other trackers are following the same period, including the University of Florida Election Lab's national tracker, which was last updated October 19, 2026, at 5:45 p.m. Eastern Time. The practical consequence is that totals move quickly and can look very different from one update to the next.
According to data compiled by callthemap.com for the period through roughly October 15 to 18, 2026, North Carolina recorded 526,000 ballots, about 37 percent above the same point in 2022. Maine recorded 97,000 ballots as of October 16, 2026, a 38 percent increase over the comparable 2022 point. Both states also showed wider Democratic registration advantages than in 2022, with North Carolina at roughly D+26.1 versus D+11.8 and Maine at D+34.2 versus D+33.6. These gaps reflect the registration of people who have voted so far, not final outcomes.
The same callthemap.com compilation shows Iowa at 33,000 ballots on day two of voting, up 61 percent versus the comparable moment, with a Democratic registration gap of about D+23.9 compared with D+10.4 in 2022. Florida, by contrast, recorded 834,000 ballots, down about 28 percent, though its Democratic gap widened slightly to D+6.1 from D+4.0. Nevada's Clark County, reporting mail ballots only, showed 41,000 ballots, down 49 percent versus the same day in 2024, with a Democratic gap of D+10.5. These differences illustrate that early-vote trends are highly state-specific rather than uniform nationwide.
In Pennsylvania, the callthemap.com data covering 61 of the state's 67 counties showed about 530,000 ballots, roughly 2 percent below the comparable point in 2022. The Democratic registration gap stood at about D+45.6, narrower than the D+54.9 recorded at the same stage in 2022. That combination of a small volume decline with a shrinking partisan edge is one example of why analysts caution against reading early numbers as a preview of the final margin. Officials in Pennsylvania, Michigan, and Georgia have all reported surges in participation according to CNNBC.
Georgia surpassed one million votes cast by the evening of October 19, 2026, reaching that mark earlier than in any previous cycle, according to Atlanta News First. On the first day of in-person early voting, October 13, 2026, the state recorded 188,667 in-person ballots, about 40 percent more than the first day in 2022 and well above the 70,849 recorded in 2018. These figures come from state reporting cited by Atlanta News First and Just the News. Georgia's early-voting window and its status as a closely watched state make its pace especially notable.
In Texas, Harris County broke its own early-voting record with at least 98,200 ballots cast on the first day, October 19, 2026, according to Click2Houston. CBS News reported that some voters in Tarrant County waited roughly an hour to cast their ballots. Texas early voting runs through October 30, 2026, giving voters a defined window before Election Day. These reports describe the first day of a multi-week period, so the county's final early-vote total will be higher.
Just the News reported that Ohio had 228,470 early votes by October 13, 2026, with its in-person early vote more than doubling compared with the prior cycle. North Carolina, which opened early voting on October 15, 2026, had recorded 237,539 ballots by that date, according to AOL's coverage. The two states differ in size and partisan lean, but both fit the broader pattern of elevated participation. Their numbers also reflect different start dates, which matters when comparing states against one another.
A notable pattern across several states is that mail-ballot returns have declined while in-person early voting has grown. AOL's reporting cited a 58 percent drop in mail returns in Georgia, alongside increases of 38 percent in early ballots in North Carolina and Maine and 61 percent in Iowa. This shift changes how quickly results can be counted, because in-person ballots are typically tabulated faster than mail ballots that arrive later. The mix of voting methods is therefore as important as the raw volume.
An NBC News Decision Desk analysis conducted with the University of Pennsylvania found that Democrats are requesting fewer mail ballots than before. In North Carolina, requests among Democratic registrants fell 40 percent, compared with 18 percent among Republicans and 23 percent among independent and third-party registrants. The same analysis found that Democrats who did request mail ballots returned them faster, with 22 percent returned in 2026 versus 17 percent in 2022. That combination of fewer requests but quicker returns is a change in behavior rather than a simple increase or decrease in turnout.
The NBC News and University of Pennsylvania analysis also examined Virginia precincts. In the state's most Democratic precincts, 69 percent more mail ballots had been returned at that point compared with 2025, while the most Republican precincts returned 5 percent fewer. This is a comparison across precinct types within one state, not a national measurement. It illustrates how the same voting method can move in opposite directions depending on the community. The finding is drawn from the analysis cited by NBC News.
In Georgia, Black voters made up 35 percent of the early vote, and women outnumbered men by about 10 percentage points, according to reporting by the Atlanta Journal-Constitution. These composition figures describe the electorate that has voted so far rather than the electorate that will vote overall. Because early voters are not a random sample of all voters, a lead at this stage does not necessarily carry through to the final count. Both parties have described the Georgia data as sending mixed signals.
The partisan breakdowns reported by NBC News combine state-reported data with modeled party affiliation produced by TargetSmart. That means a portion of the numbers is estimated rather than directly observed, and the estimates depend on the model's assumptions. Different outlets may therefore report slightly different partisan splits for the same state and date. Readers should treat the registration gaps as directional indicators rather than exact measurements.
A central caveat in the current coverage is that early-vote leads are not predictive of final outcomes. Totals update daily, and party-registration breakdowns are partly modeled, so a Democratic or Republican advantage at one point can narrow, widen, or reverse as more ballots are counted. Historically, the party that dominates early voting is not always the party that wins. The appropriate use of these numbers is to understand participation patterns, not to forecast winners.
When more voters choose in-person early voting and fewer use mail, election administrators can often report results sooner, because in-person ballots are typically processed at the polling place or shortly after. Mail ballots, by contrast, may arrive close to or after Election Day and often require signature verification and other processing steps. The behavioral shift documented in the NBC News and University of Pennsylvania analysis therefore has implications for how quickly the public learns outcomes. This is a procedural consequence rather than a partisan one on its own.
Trackers such as the NBC News early-vote page, the New York Times battleground tracker, the University of Florida Election Lab, and callthemap.com present overlapping but not identical datasets. They differ in update times, in whether they include mail ballots only or both mail and in-person, and in how they model party affiliation. A careful reader checks the update timestamp and the methodology note before drawing conclusions. Comparing two trackers with different cutoffs can create apparent contradictions that are really just timing differences.
Texas early voting runs through October 30, 2026, and Michigan's statewide in-person early voting starts October 24, 2026, according to the New York Times tracker. Election Day itself is November 3, 2026. These dates matter because a state that has not yet opened in-person voting will look artificially low in national totals. When comparing states, it helps to know each state's start and end dates for early voting. Deadlines and windows vary by state and should be confirmed with local election officials.
High overall enthusiasm combined with a shift away from mail voting gives both parties reasons for caution and for optimism. Democrats have seen strong early participation in some states but fewer mail requests, while Republicans have seen narrower registration gaps in places like Pennsylvania and Florida. Neither pattern resolves into a clear national story, which is why analysts describe the signals as mixed. The AJC's Georgia reporting used that same framing for both parties in the state.
Follow a single tracker consistently and note its update timestamp rather than mixing snapshots from different days. Compare each state against its own prior cycle at the same point, since start dates and voting methods differ. Treat party-registration gaps as estimates shaped by modeling, and treat early leads as descriptions of who has voted, not forecasts. Finally, remember that Election Day is November 3, 2026, and that most of the relevant totals will keep changing until then.
