Moody +2
(R)
47%
(D)
45%
Narrow lead, within the margin of error (±2.1 points on each candidate's share).
RedBox Research · Florida · Election night
Florida, live
Real-time election results against our final poll
- Live turnout by party against our final poll's four paths
- The count by county against what we expected
- Our projection for each race, not an official result
Status: Displaying current turnout. Awaiting election results.
As of Nov 3, 5:20 PM ET66 counties reporting live, 1 county reporting dailyUpdated: 17:20:05
Based on our October 30–31 final poll of likely voters, not based on a vote count.
U.S. Senate
Our final poll
(R)
(D)
34% chance Nixon finishes ahead
Election results
Not yet counted
Ashley Moody
(R)
(R)
0%
Angie Nixon
(D)
(D)
0%
Florida reports after polls close at 8 p.m. ET.
Governor
Our final poll
Donalds +1
(R)
47%
(D)
46%
Narrow lead, within the margin of error (±2.1 points on each candidate's share).
44% chance Jolly finishes ahead
Election results
Not yet counted
Byron Donalds
(R)
(R)
0%
David Jolly
(D)
(D)
0%
Florida reports after polls close at 8 p.m. ET.
The chance comes from our final poll's shares and its locked error model (sampling and non-sampling error together), before any vote is counted.
The party gap in turnout, now and for each possible path.
Currently R+8
Senate: Nixon +1Governor: Jolly +2
Democrats' surge grows on Election DayElection Day: R+18
Senate: Moody, Nixon evenGovernor: Jolly +1
Democrats' surge keeps going on Election DayElection Day: R+23
Senate: Moody +2Governor: Donalds +1
Our estimateElection Day: R+32
Senate: Moody +4Governor: Donalds +3
Election Day runs heavier RepublicanElection Day: R+41
Where the races turn: if voters split the way our poll found, Nixon wins with an Election Day of about R+21 or less, and Jolly with about R+28 or less. Our poll expects about R+32; Election Day 2022 ran R+27. Check-ins so far run R+25 (2,015,461 voters); the day's early check-ins can differ from its total.
The first results: to be on pace to win, Nixon needs about 56% of the early and mail votes for the two main candidates, a 13-point lead, and Jolly about 56%, a 12-point lead. Our poll expects Nixon to lead by 9 and Jolly by 10. Read them once the Panhandle reports, around 8:30 p.m. Eastern.
All figures here are Republican minus Democratic ballots: registered-party makeup, not candidate margins. The line each path starts from is the ballots cast so far, and each path ends at the makeup of the final electorate, after Election Day; Election Day under a path is what that day's own voters must run, in points of their ballots, for the final electorate to land on it. Statewide so far: R 3,230,489, D 2,594,004, NPA 1,499,021, minor 206,056.
Each county's reported Nixon share against what our poll expected of the vote counted so far.
Reported minus expected Democratic share, points (Republican side first)
- R +10 or more
- R +5 to 10
- R +2 to 5
- R +0 to 2
- D +0 to 2
- D +2 to 5
- D +5 to 10
- D +10 or more
- Not yet reported
| As of | |||||
|---|---|---|---|---|---|
| Alachua | 63% | — | — | 0% | — |
| Baker | 26% | — | — | 0% | — |
| Bay | 34% | — | — | 0% | — |
| Bradford | 32% | — | — | 0% | — |
| Brevard | 43% | — | — | 0% | — |
| Broward | 64% | — | — | 0% | — |
| Calhoun | 33% | — | — | 0% | — |
| Charlotte | 38% | — | — | 0% | — |
| Citrus | 34% | — | — | 0% | — |
| Clay | 35% | — | — | 0% | — |
| Collier | 38% | — | — | 0% | — |
| Columbia | 35% | — | — | 0% | — |
| DeSoto | 40% | — | — | 0% | — |
| Dixie | 27% | — | — | 0% | — |
| Duval | 53% | — | — | 0% | — |
| Escambia | 43% | — | — | 0% | — |
| Flagler | 41% | — | — | 0% | — |
| Franklin | 42% | — | — | 0% | — |
| Gadsden | 72% | — | — | 0% | — |
| Gilchrist | 26% | — | — | 0% | — |
| Glades | 34% | — | — | 0% | — |
| Gulf | 31% | — | — | 0% | — |
| Hamilton | 42% | — | — | 0% | — |
| Hardee | 34% | — | — | 0% | — |
| Hendry | 44% | — | — | 0% | — |
| Hernando | 40% | — | — | 0% | — |
| Highlands | 37% | — | — | 0% | — |
| Hillsborough | 54% | — | — | 0% | — |
| Holmes | 24% | — | — | 0% | — |
| Indian River | 40% | — | — | 0% | — |
| Jackson | 38% | — | — | 0% | — |
| Jefferson | 47% | — | — | 0% | — |
| Lafayette | 22% | — | — | 0% | — |
| Lake | 43% | — | — | 0% | — |
| Lee | 42% | — | — | 0% | — |
| Leon | 63% | — | — | 0% | — |
| Levy | 34% | — | — | 0% | — |
| Liberty | 32% | — | — | 0% | — |
| Madison | 45% | — | — | 0% | — |
| Manatee | 43% | — | — | 0% | — |
| Marion | 41% | — | — | 0% | — |
| Martin | 38% | — | — | 0% | — |
| Miami-Dade | 53% | — | — | 0% | — |
| Monroe | 45% | — | — | 0% | — |
| Nassau | 33% | — | — | 0% | — |
| Okaloosa | 32% | — | — | 0% | — |
| Okeechobee | 34% | — | — | 0% | — |
| Orange | 59% | — | — | 0% | — |
| Osceola | 59% | — | — | 0% | — |
| Palm Beach | 56% | — | — | 0% | — |
| Pasco | 45% | — | — | 0% | — |
| Pinellas | 52% | — | — | 0% | — |
| Polk | 49% | — | — | 0% | — |
| Putnam | 36% | — | — | 0% | — |
| Santa Rosa | 29% | — | — | 0% | — |
| Sarasota | 43% | — | — | 0% | — |
| Seminole | 50% | — | — | 0% | — |
| St. Johns | 38% | — | — | 0% | — |
| St. Lucie | 51% | — | — | 0% | — |
| Sumter | 35% | — | — | 0% | — |
| Suwannee | 31% | — | — | 0% | — |
| Taylor | 33% | — | — | 0% | — |
| Union | 28% | — | — | 0% | — |
| Volusia | 45% | — | — | 0% | — |
| Wakulla | 39% | — | — | 0% | — |
| Walton | 27% | — | — | 0% | — |
| Washington | 28% | — | — | 0% | — |
Each county's reported Jolly share against what our poll expected of the vote counted so far.
Reported minus expected Democratic share, points (Republican side first)
- R +10 or more
- R +5 to 10
- R +2 to 5
- R +0 to 2
- D +0 to 2
- D +2 to 5
- D +5 to 10
- D +10 or more
- Not yet reported
| As of | |||||
|---|---|---|---|---|---|
| Alachua | 63% | — | — | 0% | — |
| Baker | 27% | — | — | 0% | — |
| Bay | 35% | — | — | 0% | — |
| Bradford | 33% | — | — | 0% | — |
| Brevard | 44% | — | — | 0% | — |
| Broward | 65% | — | — | 0% | — |
| Calhoun | 34% | — | — | 0% | — |
| Charlotte | 39% | — | — | 0% | — |
| Citrus | 35% | — | — | 0% | — |
| Clay | 36% | — | — | 0% | — |
| Collier | 39% | — | — | 0% | — |
| Columbia | 36% | — | — | 0% | — |
| DeSoto | 41% | — | — | 0% | — |
| Dixie | 28% | — | — | 0% | — |
| Duval | 54% | — | — | 0% | — |
| Escambia | 44% | — | — | 0% | — |
| Flagler | 42% | — | — | 0% | — |
| Franklin | 43% | — | — | 0% | — |
| Gadsden | 72% | — | — | 0% | — |
| Gilchrist | 27% | — | — | 0% | — |
| Glades | 35% | — | — | 0% | — |
| Gulf | 32% | — | — | 0% | — |
| Hamilton | 43% | — | — | 0% | — |
| Hardee | 35% | — | — | 0% | — |
| Hendry | 45% | — | — | 0% | — |
| Hernando | 40% | — | — | 0% | — |
| Highlands | 37% | — | — | 0% | — |
| Hillsborough | 55% | — | — | 0% | — |
| Holmes | 25% | — | — | 0% | — |
| Indian River | 40% | — | — | 0% | — |
| Jackson | 39% | — | — | 0% | — |
| Jefferson | 48% | — | — | 0% | — |
| Lafayette | 23% | — | — | 0% | — |
| Lake | 43% | — | — | 0% | — |
| Lee | 43% | — | — | 0% | — |
| Leon | 63% | — | — | 0% | — |
| Levy | 35% | — | — | 0% | — |
| Liberty | 33% | — | — | 0% | — |
| Madison | 46% | — | — | 0% | — |
| Manatee | 44% | — | — | 0% | — |
| Marion | 42% | — | — | 0% | — |
| Martin | 39% | — | — | 0% | — |
| Miami-Dade | 54% | — | — | 0% | — |
| Monroe | 46% | — | — | 0% | — |
| Nassau | 34% | — | — | 0% | — |
| Okaloosa | 33% | — | — | 0% | — |
| Okeechobee | 35% | — | — | 0% | — |
| Orange | 60% | — | — | 0% | — |
| Osceola | 59% | — | — | 0% | — |
| Palm Beach | 56% | — | — | 0% | — |
| Pasco | 46% | — | — | 0% | — |
| Pinellas | 53% | — | — | 0% | — |
| Polk | 50% | — | — | 0% | — |
| Putnam | 37% | — | — | 0% | — |
| Santa Rosa | 30% | — | — | 0% | — |
| Sarasota | 44% | — | — | 0% | — |
| Seminole | 51% | — | — | 0% | — |
| St. Johns | 38% | — | — | 0% | — |
| St. Lucie | 52% | — | — | 0% | — |
| Sumter | 36% | — | — | 0% | — |
| Suwannee | 32% | — | — | 0% | — |
| Taylor | 34% | — | — | 0% | — |
| Union | 29% | — | — | 0% | — |
| Volusia | 46% | — | — | 0% | — |
| Wakulla | 40% | — | — | 0% | — |
| Walton | 28% | — | — | 0% | — |
| Washington | 30% | — | — | 0% | — |
Live results feed not yet available; showing turnout only.
Central-time panhandle counties (Bay, Calhoun, Escambia, Holmes, Jackson, Okaloosa, Santa Rosa, Walton, Washington; part of Gulf) vote until 8 PM ET.
Projections are ours, not official results. Official results come from the Florida Division of Elections and the county supervisors of elections; our feed reads their postings, and our turnout read comes from the counties' ballot reports by party and method (mail, early, Election Day).
Election Day lines. The Election Day lines use one measure: registered Republicans minus registered Democrats among Election Day check-ins, in points of the day's voters. Each race's line is the Election Day turnout at which it ties, given the ballots cast before Election Day, with each party's voters splitting as our final poll found; it falls between two of the turnout paths our poll modeled (“Democrats' surge grows on Election Day”, “Democrats' surge keeps going on Election Day”, “Our estimate” and “Election Day runs heavier Republican”) and is taken between them in proportion, since the races' margins move in a near-straight line from one path to the next. The first-results line is the share of the early and mail votes that ties each race if the votes counted after them split the way our poll projects. These are lines to read the night by, not forecasts.
Before results. The standing call is the poll's own: the chance the Democrat finishes ahead and the margin, from our final poll's shares and its locked error model, before any vote is counted.
Expected share. For each county and vote block (mail, early, Election Day) the expected Democratic two-party share is our poll's within-party candidate shares for that voting status (already voted for mail and early, not yet voted for Election Day) applied to the block's party mix: the county's live ballots cast by party for mail and early, the projected remaining party mix for Election Day. For the 16 benchmark counties, the mail-plus-early expectation is the region-blended first-drop benchmark.
Expected vote. Ballots already cast (live turnout by county, party and method) plus the projected remaining Election Day vote: live Election Day check-ins where a county reports them, otherwise the county's share of the remaining ballots on the “Our estimate” path, with that path's remaining party mix.
Anchor. Before results, every county's expected share moves by one statewide amount so the starting margin equals the standing call; the counted vote then moves it.
Difference from our expectation. When a county reports, its difference from our expectation is the reported Democratic two-party share minus the expected share of the blocks it has reported (mail and early count first when the vote type is not given). The rest of that county's vote is projected with that difference pulled toward its region's pooled difference by the share of the county's expected vote counted.
Uncertainty. 20,000 simulations of the uncounted vote: a statewide swing shared by every county, a regional swing and county noise, each scaled by the square root of the uncounted share. The statewide swing's typical error is set so that, before any vote is counted, the needle's chance matches the standing call (our final poll's sampling error plus the locked error model's error beyond sampling). The regional swing's typical error is half, and the county noise's all, of the gap between the race's margins on the “Democrats' surge keeps going on Election Day” and “Election Day runs heavier Republican” paths, taken as an 80% interval. Outputs: the chance the Democrat finishes ahead, the projected final two-party margin (Democrat minus Republican) with its 80% band, and the share of expected vote counted. There is no needle until real results exist.
Every figure is a whole number, rounded once from the feed's unrounded values. The page refreshes itself every 60 seconds from its own feed route, which passes Pollster's document through unchanged.
Where these numbers come from: 67 county sources, as of Nov 3, 5:10 PM ET
Each county's ballots cast by party come from its own posting, linked here with the time our feed last read it.
- Alachua · Nov 3, 5:00 PM ET
- Baker · Nov 3, 5:10 PM ET
- Bay · Nov 3, 5:00 PM ET
- Bradford · Nov 3, 5:00 PM ET
- Brevard · Nov 3, 5:00 PM ET
- Broward · Nov 3, 11:02 AM ET
- Calhoun · Nov 3, 5:10 PM ET
- Charlotte · Nov 3, 5:00 PM ET
- Citrus · Nov 3, 5:10 PM ET
- Clay · Nov 3, 5:00 PM ET
- Collier · Nov 3, 5:10 PM ET
- Columbia · Nov 3, 5:10 PM ET
- DeSoto · Nov 3, 5:10 PM ET
- Dixie · Nov 3, 5:00 PM ET
- Duval · Nov 3, 5:00 PM ET
- Escambia · Nov 3, 5:00 PM ET
- Flagler · Nov 3, 5:00 PM ET
- Franklin · Nov 3, 5:10 PM ET
- Gadsden · Nov 3, 5:10 PM ET
- Gilchrist · Nov 3, 5:00 PM ET
- Glades · Nov 3, 5:10 PM ET
- Gulf · Nov 3, 5:10 PM ET
- Hamilton · Nov 3, 5:10 PM ET
- Hardee · Nov 3, 5:10 PM ET
- Hendry · Nov 3, 5:00 PM ET
- Hernando · Nov 3, 5:10 PM ET
- Highlands · Nov 3, 5:10 PM ET
- Hillsborough · Nov 3, 5:00 PM ET
- Holmes · Nov 3, 5:10 PM ET
- Indian River · Nov 3, 5:10 PM ET
- Jackson · Nov 3, 5:00 PM ET
- Jefferson · Nov 3, 5:10 PM ET
- Lafayette · Nov 3, 5:10 PM ET
- Lake · Nov 3, 5:00 PM ET
- Lee · Nov 3, 5:00 PM ET
- Leon · Nov 3, 4:40 PM ET
- Levy · Nov 3, 5:10 PM ET
- Liberty · Nov 3, 5:10 PM ET
- Madison · Nov 3, 5:10 PM ET
- Manatee · Nov 3, 5:10 PM ET
- Marion · Nov 3, 5:00 PM ET
- Martin · Nov 3, 5:00 PM ET
- Miami-Dade · Nov 3, 5:00 PM ET
- Monroe · Nov 3, 5:10 PM ET
- Nassau · Nov 3, 5:10 PM ET
- Okaloosa · Nov 3, 5:10 PM ET
- Okeechobee · Nov 3, 5:10 PM ET
- Orange · Nov 3, 5:10 PM ET
- Osceola · Nov 3, 5:10 PM ET
- Palm Beach · Nov 3, 5:10 PM ET
- Pasco · Nov 3, 5:10 PM ET
- Pinellas · Nov 3, 5:10 PM ET
- Polk · Nov 3, 5:00 PM ET
- Putnam · Nov 3, 5:10 PM ET
- Santa Rosa · Nov 3, 5:10 PM ET
- Sarasota · Nov 3, 5:10 PM ET
- Seminole · Nov 3, 5:01 PM ET
- St. Johns · Nov 3, 5:10 PM ET
- St. Lucie · Nov 3, 5:10 PM ET
- Sumter · Nov 3, 5:10 PM ET
- Suwannee · Nov 3, 5:10 PM ET
- Taylor · Nov 3, 5:10 PM ET
- Union · Nov 3, 5:10 PM ET
- Volusia · Nov 3, 5:10 PM ET
- Wakulla · Nov 3, 5:00 PM ET
- Walton · Nov 3, 5:10 PM ET
- Washington · Nov 3, 5:00 PM ET
The model's technical note
It uses the model's own names: N is “Democrats' surge grows on Election Day”, P0 is “Democrats' surge keeps going on Election Day”, midpoint is “Our estimate”, B1 is “Election Day runs heavier Republican”, a residual is the difference from our expectation and sd is the typical error (standard deviation).
Florida 2026 election-night needle: how it works.
Inputs (config/florida_2026_election_night_inputs.json, each source with its sha256): our final poll's within-party
candidate shares by voting status (already voted / not yet voted, approved weights), the region-blended first-drop
expectations of the 16 benchmark counties, and the three final-turnout paths (P0 / midpoint / B1).
1. Expected share. For each county and vote block (mail, early, Election Day) the expected Democratic (Nixon; Jolly
for Governor) two-party share is the poll's within-party shares for that voting status times the block's party mix:
mail and early use the already-voted shares and the county's live ballots cast by party; Election Day uses the
not-yet-voted shares and the county's own remaining party mix. A benchmark county's mail and early blocks also use
its live banked mix; its region-blended benchmark enters only as an adjustment, the benchmark value minus the
poll's expectation at the benchmark's own banked mix.
2. Expected vote. Each county's expected final ballots are the midpoint turnout path's (inputs expected_vote, by
party); the Election Day block is live Election Day check-ins plus the remaining ballots, final minus cast so far
by party, floored at 0, so check-ins are a cumulative lower bound and never a sudden final. The counted share and
every scaling use these totals; the P0 and B1 paths give the scenario expected margins.
3. Residuals. When a county reports, residual = reported Democratic two-party share minus the expected share of the
blocks it has reported. Residuals are shrunk toward 0 by their precision (reported votes), pooled by region without
the county itself, capped when transferred, and carried into the Election Day block at 0.6 until Election Day votes
report. A county reporting totals with no vote-type split has its read damped by its unknown block composition:
the spread of the expected share of its reported votes between filling mail first, early first, and all blocks in
proportion. Until the Central-time panhandle's polls close (8 PM ET), no residual is transferred into those
counties' Election Day vote (half for Gulf, which is split).
4. Anchor. Before results, the expected statewide margin is the outcome call's (services.outcome_call on the approved
read and the locked W3 error file, inputs "prior"): the call-vs-turnout gap is carried in the Election Day block
only, by an offset frozen in the inputs (prior.<race>.anchor_offset_ed, computed once from the inputs' own banked
snapshot and recorded with the inputs sha256), so later turnout moves the expectation instead of being cancelled.
5. Uncertainty. 20,000 simulations of the uncounted vote: a statewide swing shared by every county (its posterior sd
given the reporting counties), an Election Day block swing (sd floor 2.75 share points until Election Day votes
report), a regional swing (six poll regions) scaled by the square root of the region's uncounted share, and county
noise; each county's Election Day vote total also varies, log-scale sd its P0-to-B1 Election Day range taken as an
80% interval. Regional sd = half and county sd = all of the P0-to-B1 path range taken as an 80% interval,
|margin(P0) - margin(B1)| / (2 * 1.2816), in two-party points. The statewide sd is set so that, with nothing
counted, the simulated statewide two-party margin's sd equals the outcome call's (less the local and Election Day
parts). The call's sd is sqrt(sampling^2 + W3 nonsampling^2) in all-respondent D - R points; the needle works in
two-party points 100 * (D - R) / (D + R), so the call's margin and sd are each divided by the poll's D + R share.
6. Outputs. P(Democrat finishes ahead), projected final two-party margin (D - R, points; the simulations' median)
and its 80% band, and the
share of expected vote counted. No needle until real results exist (phase "turnout"); races.*.prior carries the
call (p_dem_ahead, margin, sd in two-party points; margin_share and sd_share in all-respondent points) and the
simulation's own P with nothing counted.