Beta

About this page

How OhioVotesCount.com's election-night page works: where the numbers come from, what the charts and projections show, and what each note means — and doesn't mean.

This page is in beta. November 3, 2026 is its first use on a live election. It was tested on the full 2024 Ohio results, on recent primary-election reports and on simulated data, but problems only a live count can reveal are possible. All numbers are unofficial election-night counts. Official results come from the Ohio Secretary of State after the counties certify them, about two weeks after the election.

1. What the page shows

For every Ohio contest on the ballot that the New York Times reports — statewide offices, U.S. Senate, the 15 congressional districts, the Ohio Supreme Court, the General Assembly districts and statewide issues — you can pick the contest and see how each candidate's (or For/Against's) share of the counted vote changed over the evening, with each candidate's current vote total in the box above the chart and a table of the latest totals.

The page checks for new numbers every 30 seconds by itself; there is no need to reload it. The underlying data is recorded once a minute.

2. Where the numbers come from

Ohio's 88 county boards of elections count the ballots. A board's unofficial results then reach the public by three routes, and this page reads all three. Throughout the page, each is called by the name in bold:

The county's own results: the results the county board of elections posts itself, straight from its tabulation system, as a report (often a PDF) or a results page on its website. Some counties post through a results site run for them by a vendor (EnhancedVoting); that is still the county's own posting.
The Secretary of State's results site: the Ohio Secretary of State's election-night results site (liveresults.ohiosos.gov), which each county uploads to on a schedule (every 15, 30 or 60 minutes, depending on the county). These are election-night figures too, not the certified official results.
The NYT feed: the New York Times election results feed. Its vote counts come from the Associated Press (AP), whose reporters at the boards and staff reading the state and county sites collect counts county by county; the feed itself names AP as its source. (It is sometimes called the "NYT/Edison feed": Edison Research supplies counts to some other news organizations, but for Ohio the NYT feed credits AP. The page footer shows the source the feed names.)

What comes from where. By default the chart and the totals table, and always the county map and the projected final shares, come from the NYT feed, recorded once a minute. The Secretary of State's results site (read every two minutes) and the county's own results (read every five minutes, for the counties whose results can be read automatically) are used for the notes, and each note below says which source or comparison it was detected in. The chart's Source switch can also draw the chart and table from either of them (see below). Two notes also use other data: the "more votes than registered voters" note uses the registered-voter counts in the Ohio Secretary of State's statewide voter file, as of its most recent monthly release before the election, and the pattern and turnout notes use each county's final 2024 results from the NYT feed and its certified 2022 results from the Ohio Secretary of State.

Every version of every file we read is archived as it was received, so any number shown here can be traced back to its source.

3. Reading the chart

One contest or four. "One contest" shows a large chart with the full table, notes and details. "Four contests" shows four smaller live charts at once, two over two, each with its own contest picker (on a phone they stack). Your choices are kept in the page address, so a bookmark or a shared link opens the same contests.
Share of vote or vote totals. "Share of vote" (the default) shows each candidate's share of the votes counted so far; "Vote totals" shows each candidate's cumulative votes. The switch applies to every chart on the page. The current totals are always in the box above each chart.
Source: NYT/AP, Secretary of State or Ohio counties. The same chart, drawn from a different feed. "Secretary of State" adds up that site's figures for every county in the contest; "Ohio counties" adds up the counties' own results. Each county counts at its latest posted total until it posts a new one. A county whose own results can't be read (or haven't been posted) is left out of the "Ohio counties" line, not filled in from another source: the chart says how many counties are included and names the ones that aren't, and until every county is included that line runs below the others. Once every county is in, all three lines should end up at the same totals; they step at different times because each source updates on its own schedule. The projection is drawn only on the NYT/AP view, because it uses AP's estimate of the expected vote. The switch is off for a contest the Secretary of State's site has no match for.
Lines move in steps: each step is a reported update, and the line holds until the next one.
The time axis runs from 7:30 PM, when Ohio's polls close, to 11:59 PM, and later if counties are still reporting large batches (up to 8 AM). Use "Everything" to see the later updates as provisional and late overseas ballots are added in the following weeks.
Early swings in the shares are normal. Ohio counties report early and mail ballots first, starting at 7:30 PM, and Election Day votes after that. These groups often lean differently, so the shares can move a lot during the evening without anything being wrong. (The first absentee results are due by 8 PM.)
Hovering over the chart shows each update's time, the share of the expected vote counted, and each candidate's share and votes. An update marked "from NYT's published history" is one our recorder didn't see itself (for example while it was briefly offline) and was filled in from the history NYT publishes.
The county map ("Counted so far, by county", below the table in the one-contest view) shades each county by the share of its expected vote counted, from the NYT feed (the expected vote is AP's estimate): the darker, the more counted (in dark mode, the brighter). Hover over a county, or tap or click it, for its percentage and votes counted. For a congressional district, the counties outside the district have dashed outlines. NYT reports State Senate and State House districts as district totals only, so they have no map. The ▶ button replays the map from the first returns to now, like a weather radar loop, in about 12 seconds; the slider moves to any moment, and the time it shows is when those counts were reported. Each county shows its latest count as of that moment.
"Vote count went down" (a red box under the chart) lists every time a candidate's statewide total dropped from one update to the next, with the county where it happened and that county's count before and after. NYT's data gives the statewide and county totals of each update together, so the drop can be traced to the county whose own total fell at the same moment. If no county's total fell, the box says the statewide figure itself was revised. Totals are cumulative, so a drop means a number was changed after it was initially reported: possibly a county or AP fixing an entry error. It is worth a look, not evidence of a problem by itself.

4. The status indicator

Live · updated (time): the page is receiving results; the time is the last change in any contest.
Waiting for the first results: the New York Times has not published Ohio's results yet.
NYT results feed unreachable, retrying: our recorder could not read the NYT feed on its last try.
Updates paused: nothing has been recorded for more than 10 minutes because our recorder is not running. The numbers shown were right as of that time. When it resumes, the latest numbers return at once; for the main statewide races and issues, the updates missed in between are filled in from the history NYT publishes.

5. Projected final shares

Once about a quarter of a contest's expected vote has been counted, the chart may extend each line with a dashed line to its projected final share (or, with "Vote totals", its projected final vote), shaded to show a range, and the table adds a "Projected final share" column.

What it is: arithmetic on the counts reported so far in the NYT feed. For each county, the votes still expected (AP's estimate) are added to what has been counted, split two ways: like that county's votes so far, and like its most recent updates. The projection is the middle of the two; the shaded range spans them. As a share, the line bends because each added vote moves a cumulative share less as the total grows; as a vote total it is straight.

What it is not: a forecast model, a poll or a race call. It knows nothing about which kinds of ballots are still uncounted. Early in the count it can be far off, even about who is ahead — replayed on 2024's results, at about 25% counted it showed the wrong leader in the presidential race and on Issue 1, because the early and mail ballots counted first leaned the other way; by about half counted it had the right leader in every statewide contest. Final results can fall outside the shaded range. The expected vote is itself an estimate and was about 5% too high in 2024.

6. Noteworthy counts: every note explained

Below the chart, a contest may list noteworthy counts: things in the county-by-county numbers that deserve a second look. A note is not a finding of error or wrongdoing. Each one comes with its common innocent explanations, and most notes on most nights will turn out to be timing, corrections or data entry. Notes are labeled Look into, Noteworthy or Minor by how unusual the pattern is, not by how likely something is wrong. A note stays listed after the numbers are corrected, marked "No longer the case", so the record of what happened is kept.

Following up a note. Each note says where it was detected (the NYT feed, the Secretary of State's results site, the county's own results, or a comparison of two of them), names the county where the change happened whenever it can be traced, gives the counts involved and their times, and ends with Check it links: NYT's page for the race, the Ohio Secretary of State's results site and the county's own results (or its board of elections). Every file we read is archived as received, and when a note first appears, the data from every source at that moment is saved together, with a fingerprint of each file to show later that nothing was changed. The note's evidence reference names that record; quote it to OhioVotesCount.com to ask for the data behind a note.

Checks on the NYT feed's numbers

These look only at the numbers in the NYT feed (AP's vote counts): each contest's total and its county totals.

More votes than registered voters Look into

Detected in: The NYT feed, county totals, compared with the registered voters in the Secretary of State's statewide voter file.
Means: A county's total in a contest is higher than the number of registered voters in that county.
Doesn't mean: Fraud, or extra ballots. The most common causes are an entry error (a digit added, a batch entered twice) or votes recorded under the wrong county or contest. Our registration count comes from the latest monthly statewide voter file, which can lag slightly behind the county's own rolls.

Counties don't add up to the total Look into

Detected in: The NYT feed: its statewide or district total compared with its own county totals, in the same file.
Means: In the same file, a candidate's statewide total differs from the sum of the 88 counties, or a congressional district's total from the sum of the counties in that district, by more than 10 votes and 0.05%, in two readings in a row. (For a county split between districts, NYT reports only the part of it in the district.)
Doesn't mean: That votes were added or lost. The statewide and county figures may have been updated a moment apart, some ballots may be counted statewide before being assigned to a county, or a correction may have reached one level first.

Count went down Noteworthy

Detected in: The NYT feed: the contest's statewide or district total, and its county totals.
Means: A candidate's statewide (or district) total was lower than in the previous update. The note names the county or counties whose own totals fell in the same update, with their counts before and after.
Doesn't mean: Votes being taken away. A possible cause is a county or AP fixing an entry error. If no county is named, AP or NYT revised the total itself.

County count went down in the NYT feed Noteworthy

Detected in: The NYT feed, county totals. (The same kind of drop on the Secretary of State's results site is a separate note, below.)
Means: A candidate's total in one county was lower than in that county's previous update.
Doesn't mean: Votes being taken away. Cumulative totals only go down when a number is changed after it was first reported: possibly an entry error fixed, precinct results reported twice and then removed, or ballots moved to the right contest after review.

Late shift in a county's vote share Noteworthy Look into

Detected in: The NYT feed, county totals.
Means: After a county had counted about 40% of its expected vote, a single update moved a candidate's share of that county's vote by 1.5 points or more. It is labeled "Look into" if the update was also larger than the vote the county was expected to have left.
Doesn't mean: Something unusual happened to those ballots. Later batches are often different: Election Day votes arriving after the early and mail ballots (the most common reason before about 80% counted), a city or group of suburbs reporting last, absentee or provisional ballots processed late, or a correction. The expected-vote estimate itself can be too low, making a county look more complete than it was. This check deliberately ignores the first part of a county's count, when large swings are routine.

More votes down the ballot than in the top race Minor

Detected in: The NYT feed, county totals of two contests in the same file.
Means: In a county, a statewide race or issue further down the ballot has more votes than the top race (Governor this year), by more than 25 votes and 0.5%, in two readings in a row.
Doesn't mean: Extra ballots. Usually fewer voters mark the later races, but an issue can draw voters who skip the top race, the two contests may have updated at different moments, or votes may have been entered under the wrong contest.

The same batch of votes added twice Noteworthy Look into

Detected in: The NYT feed, each county's successive updates. The same check also runs on the Secretary of State's results site, reading by reading; a note found there says "on the Secretary of State's site" and names it as the source.
Means: An update to a county's count added exactly the same votes for every candidate as an earlier update (for example +312 and +188, twice), in batches of at least 100 votes; or two counties added exactly the same votes at the same moment. It is labeled "Look into" when the same thing happens at the same updates in two or more contests in the county, which is what results entered twice look like. Two updates in a row that are only nearly identical (every candidate within 0.5%, 500 votes or more) are listed as Minor.
Doesn't mean: That votes were counted twice. A correction can remove results and add them back, and nearly identical batches occur by chance now and then. Identical batches of 100 votes or more are rare by chance: in simulated election nights built from 2024's county results, fewer than one in ten nights showed one in any contest. It does not catch results doubled within a single update; the registered-voter, turnout and three-source checks cover that case.

County result far from its usual pattern Noteworthy Look into

Detected in: The NYT feed, county totals, compared with the county's final 2024 U.S. Senate result from the same feed.
Means: Once a county has counted about 70% of its expected vote, its share for the Democratic candidate, out of the votes for the two main candidates, is at least 6 points away from what its past voting predicts. The prediction is the county's 2024 U.S. Senate share plus the change seen this year in the counties at a similar stage of their count. Counties keep their lean from one election to the next: in 2024, one statewide race predicted another within about half a point in the typical county and 3 points at worst. The limit is 6 points, or four times the typical miss among the other counties if that is larger; twice the limit is labeled "Look into". Statewide partisan contests only.
Doesn't mean: That the count is wrong. A county can genuinely vote differently this year (a local candidate, a local issue, a shift in who turned out), and its votes counted so far may be more of one kind (early, mail or Election Day) than in other counties. It is a pointer to where an entry error would show up, such as votes entered under the wrong candidate.

Unusual turnout in a county Noteworthy

Detected in: The NYT feed, county totals in the top race, compared with the county's final 2024 presidential vote from the same feed.
Means: A county's votes so far in the top race are more than 10% higher than all the votes it cast for President in 2024 (a midterm usually draws about three quarters as many), or, once most counties are 95% counted, the county's turnout compared with 2024 is far out of line with the other counties (15 points or more from the typical county).
Doesn't mean: Extra ballots. Growing counties and local races or issues can raise turnout; the comparison with 2024 does not fit every county. It can also point to votes entered twice or under the wrong county.

More votes than in the county's 2022 midterm Noteworthy

Detected in: The NYT feed, county totals in the top race (Governor), compared with the county's official 2022 votes for the same office, from the Ohio Secretary of State's certified 2022 results.
Means: A county's votes so far for Governor are more than 10% higher than all the votes it cast for Governor in 2022, the last midterm. Only in a midterm year, when the two elections are alike.
Doesn't mean: Extra ballots. Turnout can rise from one midterm to the next (a closer statewide race, a local issue), and fast-growing counties gain voters; if turnout is up across the state, several counties can pass the limit together. It can also point to votes entered twice or under the wrong county.

Reporting delays

These watch how often each source updates, from polls closing at 7:30 PM until 3 AM, in the top race on the ballot (every county reports it): the NYT feed, the Secretary of State's results site, and the counties' own results we read. Each note says which source went quiet; comparing the other two shows whether the county or only that source stopped. A delay counts only while we kept reading the source successfully and its results didn't change, so a source we couldn't reach is never reported as a county or a source that stopped.

A county stopped reporting before it finished counting Noteworthy

Detected in: One source's county counts (the note says which), compared with the same source's other counties.
Means: No new count from the county for 90 minutes or more while it had counted less than 95% of its expected vote (AP's estimate) and the same source kept updating other counties. On the NYT feed and the Secretary of State's site, also a county with no results at all 90 minutes after polls closed. The note stays after the county resumes, with when its next count came.
Doesn't mean: Something is wrong. Counties often wait for ballots or memory cards from outlying precincts, or count a large batch before their next upload, and some stop for the night with ballots left to count. If only one source is behind, the delay is in that source (AP's reporter, the county's upload to the Secretary of State, or the county's website), not in the county's count.

A source stopped updating Look into

Detected in: One source's county counts (the note says which).
Means: No new count from any county on that source for 30 minutes or more while counties were still counting (90 minutes right after polls close, before the first results).
Doesn't mean: Results are being held back. A results site or feed can freeze on old figures, and late at night every county can be between uploads at once. The other two sources show whether the counties themselves stopped.

When we can't read a source

Separate from the results: these say when this monitor itself couldn't get a source's results, so a gap in the checks is never mistaken for a quiet night.

We couldn't read a source Look into

Detected in: This monitor's own record of reading each source.
Means: No successful read of the NYT feed for 5 minutes, of the Secretary of State's results site for 10, or of a county's own results for 20 (from 9 PM, when every county should have posted), with the last error we got. While it lasts, that source's counts are missing from the checks. The note stays after reading resumes.
Doesn't mean: Anything about the results. The problem can be on our end (the internet connection, the computer running the monitor) or the source's site being down, slow or blocking requests.

Counties whose own results we aren't reading Minor

Detected in: This monitor's list of the county results it reads.
Means: The counties whose own results aren't part of the checks: no results address known for them, or results posted as a scanned image with no numbers to read.
Doesn't mean: Those counties go unchecked: their counts are still compared between the NYT feed and the Secretary of State.

Checks on the Secretary of State's results site

These look only at the county totals on the Secretary of State's results site (liveresults.ohiosos.gov). It is the only county-level source we have for State Senate and State House districts, since the NYT feed reports those as district totals only.

County count went down at the Secretary of State Noteworthy

Detected in: The Secretary of State's results site, county totals, between two of our readings.
Means: A candidate's total in one county on the Secretary of State's site was lower than at our previous reading of that site (we read it every two minutes).
Doesn't mean: Votes being taken away. Cumulative totals only go down when a number is changed after it was first reported: possibly the county fixing an entry error in an earlier upload, precinct results uploaded twice and then removed, or a brief problem with the site itself (check whether the next reading goes back up).

Checks comparing the sources

These compare two sources county by county: the NYT feed with the Secretary of State's results site, or the county's own results with the Secretary of State's results site. For a State Senate or State House district that lies within one county, NYT's district total is that county's count, so it is compared exactly like a county. For a district of several counties, NYT gives only the district total, so it is compared with the Secretary of State's counties added up (below), and each county's own results are compared with the Secretary of State as usual.

NYT count never shown by the Secretary of State Look into

Detected in: The NYT feed compared with the Secretary of State's results site (and with the county's own results, where we read them).
Means: For a county and contest, NYT showed a count that, 75 minutes later, neither the Secretary of State's site nor the county's own results had ever shown. (75 minutes allows for counties that upload only once an hour.)
Doesn't mean: That the count was wrong. It is the clearest sign of a mistake somewhere between the county and the public — a transcription error by a reporter relaying numbers, a processing error — but a county can also give AP a count that it uploads to the state with different figures or later than scheduled.

County's own count never shown by the Secretary of State Look into

Detected in: The county's own results (its website or posted report) compared with the Secretary of State's results site.
Means: A county's own published count has stayed its latest for more than 75 minutes, and the Secretary of State's site has never shown that count.
Doesn't mean: A discrepancy in the ballots. The county may have updated its own site after its last scheduled upload and the next upload is late, or uploaded different figures by mistake. Our reading of the county's report could also have picked up the wrong number — every report is archived so this can be checked.

NYT and the Secretary of State still differ Noteworthy

Detected in: The NYT feed compared with the Secretary of State's results site.
Means: NYT has stayed on an older count of a county for more than 75 minutes after the Secretary of State moved on.
Doesn't mean: That one of them is wrong. One source may have stopped updating for that county, or a correction may have been made at one source but not the other.

NYT's district total differs from the Secretary of State's Noteworthy

Detected in: The NYT feed's district total compared with the Secretary of State's results site's county totals for that district, added up.
Means: For a State Senate or State House district spanning several counties, NYT's total and the sum of the Secretary of State's counties are different, and neither has changed for more than 75 minutes.
Doesn't mean: That one of them is wrong. One source may have stopped updating for one of the counties, or a correction may have been made at one source but not the other. Because NYT gives no county figures here, the note can't say which county; the Secretary of State's and the counties' own results for the district are the place to look.

Secretary of State showed NYT's count much later Minor

Detected in: The NYT feed compared with the Secretary of State's results site.
Means: The Secretary of State's site showed the same count NYT had, but more than 75 minutes later.
Doesn't mean: Anything was wrong: the numbers matched. AP's reporters at a board can relay a count before the county's scheduled upload, or an upload can be delayed.

Checks on the data changing after the fact

We keep our own record of every count as we first saw it and never overwrite it. If a source later shows something different for a moment it already reported, we note it. These checks cover the NYT feed only (its live file and the history it publishes); the Secretary of State's results site and the counties' own results publish no history to compare against.

NYT's history differs from what we recorded Look into

Detected in: The NYT feed: the history it publishes compared with what we recorded from its live file.
Means: The history NYT publishes now shows different totals for a moment than the totals we recorded live at that moment.
Doesn't mean: That anyone altered results. Corrections are sometimes applied to the stored history rather than as a new update, and updates can be re-timestamped. Our record keeps the version first published.

NYT changed a count in its published history Look into

Detected in: The NYT feed: two versions of the history it publishes.
Means: NYT's published history now shows different totals for a moment than an earlier version of that same history did.
Doesn't mean: Wrongdoing; same explanations as above. Both versions are archived.

An update disappeared from NYT's history Look into

Detected in: The NYT feed: two versions of the history it publishes.
Means: An update NYT had published is no longer in its history.
Doesn't mean: That votes disappeared. An update later found to be wrong may have been removed, or merged with another.

A count changed without a new timestamp Noteworthy

Detected in: The NYT feed's live file, compared with what we recorded from it earlier.
Means: The NYT feed changed a count but kept the update's original time.
Doesn't mean: Anything improper. A correction may keep its original time, or two updates may fall in the same second.

7. Limits and known gaps

Unofficial, election-night numbers. Provisional ballots and late overseas and military ballots are added in the following weeks; final results can differ, especially in close contests.
Coverage of the cross-checks. County-by-county checks of NYT against the Secretary of State cover the statewide contests, State Issue 3, the 15 congressional districts and the State Senate and State House districts that lie within one county. For a State Senate or State House district of several counties, NYT reports only the district total, so that total is compared with the sum of the Secretary of State's counties; it can show that the two disagree, but not in which county. Some counties' own results can't be read automatically (a few post scanned images or block automated access); for those, the check compares NYT with the Secretary of State only.
Timing. Each source updates on its own schedule, so for up to an hour or so they can legitimately show different numbers. The notes are designed to wait out that lag before flagging anything.
Write-in candidates are left out of the cross-checks; they are often reported differently by each source.
Our own errors. This is new software reading other organizations' data, which can change format without notice. If something here looks wrong, the sources linked above are authoritative.