The 21st CenturyElections Engine

2026 / Early vote · Election Day

Ballots already cast across the country, before polls even opened

—

That’s the early in-person and mail vote the states have reported so far.

As of · Source: University of Florida Election Lab (Dr. Michael McDonald), which collects state and county election-office reports. States report on their own schedules, so some are a day or two behind.

The map

Where the ballots are

The stronger the color, the more ballots. Tap a state for its count, how it compares with earlier years and, where the state reports it, the party-registration mix. Hatched states have no data yet. That’s not the same as zero.

Small states are easier to hit in Tiles view.

Compared with 2016 and on

The national early vote, cycle by cycle

Midterms are the fairest comparison, so 2018, 2022 and this year are in purple. Presidential years (2016, 2020, 2024) are grey: they always run bigger. Past years are each cycle’s last report, usually at or just after Election Day. 2026 is this morning’s count and still growing.

MidtermPresidential2026, so far today

    Older years are less complete. In 2016 and 2018 several states hadn’t reported in-person early voting when the sheet was last updated, so those two totals run low. † marks a state whose last report came a week or more before Election Day.

    Party registration

    Who is voting early, by registration

    State by state

    Every state, 2016 to 2026

    Midterm columns (2018, 2022, 2026) are shaded. “No data” means the source has nothing for that state and year. Tap a row for the full picture.

    Where these numbers come from

    All 2026 numbers come from one file, UF Election Lab’s 2026 early-vote data. The national figure is simply those state numbers added together, and the map, chart and table all read that same file. “Early vote” here means in-person early voting plus returned mail ballots, as each state defines them. Ballots aren’t votes for a candidate: nothing is counted until polls close, and party registration shows who voted, not how.

    No data for: (2026). Where states don’t split in-person from mail, we show one combined count. As of .

    Past years come from the same tracker’s earlier editions:

      Every year uses each state’s own last report, and some states changed how they count over time, so treat small gaps with care.

      Closer look

      North Carolina and Georgia: the official counts, in detail

      Our earlier tracker, straight from the two states’ election offices. It goes deeper than the national file: voting groups, history and county detail.

      2026 / EARLY VOTE

      Who isshowing up.

      Follow official reported ballots, voting groups and the election stages behind the numbers.

      Checking the official observations…

      REPORTED BALLOTSUnavailable

      Official administrative observation

      WHERE THERE ARE OFFICIAL COUNTS

      North Carolina and Georgia, up close.

      THE REPORTED PACE

      The counts so far.

      Read the observed counts and dates

      THE PARTICIPATION MIX

      The groups behind the count.

      Registered party describes registration, not whom someone voted for.

      HISTORICAL CONTEXT

      Put the election stage in view.

      THE SOURCE RECORD

      What the numbers cover.

      Original sources, dates and coverage

      Ballots are not candidate votes. Each state keeps its own cutoff and ballot status, so the national number at the top comes from the UF Election Lab file, not from adding these two states.

      Age is unavailable in the reviewed aggregates. Unknown demographic categories remain included. Requested, returned, accepted and cast ballots keep their source labels.

      SEPARATE RESEARCH

      Observed counts first.

      No final-turnout or candidate-vote prediction is issued here. Administrative early-vote counts are not an active numerical term in the inspected 2026 engine. Experimental use with polling requires separate held-out testing; this tracker changes no model coefficients or engine schedules.

      Observations keep their stated cutoff and source date. Historical editions may contain later revisions; they are descriptive evidence, not proof of what a forecaster could have known at an earlier deadline.