How many Marylanders will vote in 2026?
A forecast of ballots cast in the November 3, 2026 general election, by county and party, built from the statewide voter list, the 2018 and 2022 midterms and the State Board's 2026 mail-in and early-voting reports. Mail-in data through October 30, 2026; early voting through October 29, 2026. Model last run November 1, 2026.
The short version
- The central forecast is 2,330,865 ballots, 54.2% of active registration (50.2% of all eligible voters, active plus inactive).
- The range to quote is 2,060,722 to 2,553,256 ballots, 47.9% to 59.4% of active registration. For comparison, 2,031,635 ballots were cast in 2022 (49.3%) and 2,335,128 in 2018 (59.1%).
- Early votes plus the mail ballots expected in all come to 1,164,939, 26% more than the same count in 2022 (922,864). The central forecast needs 37.2% of the active registrants who have not yet voted to turn out on election day; in 2022 that rate was 34.6%, in 2018 48.9%.
Three scenarios
The model does not pick one turnout level. It calibrates each party's turnout to the two most recent midterms, 2022 (low) and 2018 (high), and sets a central scenario between them using a rule written before the 2026 numbers came in. The spread between low and high is the real uncertainty: the 90% intervals within each scenario are narrow because they carry only county-level noise.
| Scenario | Ballots cast | 90% interval | Rate on active | Rate on eligible |
|---|---|---|---|---|
| Low: a 2022-like midterm | 2,102,288 | 2,060,722 to 2,143,646 | 48.9% | 45.3% |
| Central: the model's blend | 2,330,865 | 2,289,592 to 2,371,828 | 54.2% | 50.2% |
| High: a 2018-like midterm | 2,513,478 | 2,473,047 to 2,553,256 | 58.4% | 54.1% |
| Actual, 2022 | 2,031,635 | 49.3% | ||
| Actual, 2018 | 2,335,128 | 59.1% |
The State Board's 2018–2024 turnout sheets divide by active registration; its 2026 guide divides by all eligible voters, active plus inactive. The 2026 eligible count is projected, so the Board's own percentage will differ slightly.
How the central scenario is set
Each party starts halfway between its 2022 and 2018 turnout. Four measures of 2026 engagement, each compared with 2022, can move it 10 points of the way toward one year or the other, between 20% and 80% of the way to 2018.
| Party | Registration momentum | Mail ballots sent per registrant | Primary turnout per registrant | Ballots cast before election day per registrant | Weight on 2018 |
|---|---|---|---|---|---|
| Democrats | toward 2022 | neither | neither | toward 2018 | 50% |
| Republicans | neither | neither | toward 2022 | toward 2018 | 50% |
| Unaffiliated and other | toward 2018 | neither | neither | toward 2018 | 70% |
Ballots already cast
604,396 people voted early (the State Board's unofficial count; early voting is over) and 376,967 mail ballots had come back by October 30, 2026. At 2022's rate of late arrival, the model expects 560,543 mail ballots in all. Same-day registration above 2022's level adds 7,142 ballots to every scenario.
| How ballots are cast (central) | Ballots | Share | 2022 share |
|---|---|---|---|
| 560,543 | 24.0% | 26.7% | |
| Early voting | 604,396 | 25.9% | 18.7% |
| Election day | 1,093,167 | 46.9% | 51.5% |
| Provisional | 72,759 | 3.1% | 3.1% |
The share of active registrants who have not yet voted that each scenario needs on election day: low 29.9%, central 37.2%, high 43.0%. The same rate was 34.6% in 2022 and 48.9% in 2018. Some of 2026's early surge is voters who used to vote on election day, so a large early vote does not by itself mean a large turnout.
Who votes
Party is the voter's registered party. The model forecasts each party's turnout separately, so the mix of the electorate follows from those rates.
| Party | Central turnout (active) | 2022 | 2018 | Central share of ballots | 2022 | 2018 | UMBC Institute of Politics likely voters |
|---|---|---|---|---|---|---|---|
| Democrats | 56.6% | 50.8% | 61.5% | 53.9% | 55.9% | 57.3% | 60% |
| Republicans | 60.7% | 57.6% | 63.5% | 26.2% | 28.1% | 27.4% | 25% |
| Unaffiliated and other | 43.1% | 36.2% | 46.4% | 19.9% | 16.0% | 15.3% | 15% |
The UMBC Institute of Politics poll (fielded through September 27, 2026) reports its likely voters' party mix unweighted; it is shown for comparison and does not enter the model.
Counties
Turnout as a share of each county's active registration. Click a column heading to sort.
| County | Active | Central ballots | Low | Central | High | 2022 | 2018 | Early | Election day | |
|---|---|---|---|---|---|---|---|---|---|---|
| Allegany | 44,772 | 24,098 | 49.3% | 53.8% | 57.5% | 50.5% | 55.8% | 15% | 15% | 68% |
| Anne Arundel | 419,882 | 238,116 | 51.4% | 56.7% | 60.7% | 53.7% | 60.1% | 28% | 27% | 42% |
| Baltimore City | 413,160 | 178,851 | 37.8% | 43.3% | 48.2% | 37.3% | 48.1% | 20% | 19% | 55% |
| Baltimore County | 586,484 | 317,742 | 48.8% | 54.2% | 58.5% | 49.3% | 59.4% | 22% | 25% | 50% |
| Calvert | 70,199 | 41,539 | 54.2% | 59.2% | 62.9% | 55.9% | 61.1% | 21% | 32% | 45% |
| Caroline | 22,911 | 12,347 | 49.3% | 53.9% | 57.6% | 50.4% | 57.5% | 11% | 25% | 63% |
| Carroll | 130,185 | 78,232 | 55.4% | 60.1% | 63.6% | 57.1% | 62.6% | 17% | 24% | 57% |
| Cecil | 73,959 | 38,814 | 47.8% | 52.5% | 56.2% | 47.5% | 54.5% | 15% | 25% | 58% |
| Charles | 130,599 | 68,420 | 46.9% | 52.4% | 56.9% | 45.5% | 58.5% | 17% | 33% | 47% |
| Dorchester | 24,272 | 13,302 | 50.0% | 54.8% | 58.7% | 51.5% | 58.9% | 15% | 25% | 58% |
| Frederick | 207,968 | 122,194 | 53.6% | 58.8% | 62.6% | 55.0% | 62.9% | 22% | 23% | 52% |
| Garrett | 21,124 | 12,195 | 53.6% | 57.7% | 61.1% | 56.3% | 57.0% | 14% | 23% | 62% |
| Harford | 198,397 | 116,626 | 53.9% | 58.8% | 62.5% | 55.2% | 61.9% | 15% | 29% | 54% |
| Howard | 240,569 | 148,756 | 56.4% | 61.8% | 65.8% | 57.4% | 67.6% | 22% | 26% | 49% |
| Kent | 14,412 | 9,264 | 59.5% | 64.3% | 67.9% | 62.0% | 67.0% | 18% | 28% | 53% |
| Montgomery | 702,193 | 400,011 | 51.3% | 57.0% | 61.3% | 51.5% | 63.0% | 35% | 23% | 37% |
| Prince George's | 623,028 | 300,118 | 42.6% | 48.2% | 53.0% | 40.6% | 55.8% | 27% | 32% | 37% |
| Queen Anne's | 41,775 | 25,758 | 57.0% | 61.7% | 65.2% | 59.0% | 64.9% | 15% | 31% | 52% |
| Somerset | 14,743 | 7,973 | 49.4% | 54.1% | 58.0% | 49.3% | 60.2% | 13% | 25% | 61% |
| St. Mary's | 78,772 | 43,273 | 50.0% | 54.9% | 58.8% | 50.5% | 58.0% | 17% | 33% | 47% |
| Talbot | 28,912 | 18,916 | 60.7% | 65.4% | 69.0% | 62.7% | 68.4% | 21% | 37% | 41% |
| Washington | 101,195 | 52,972 | 47.5% | 52.3% | 56.1% | 49.9% | 54.2% | 18% | 25% | 55% |
| Wicomico | 67,383 | 34,537 | 46.3% | 51.3% | 55.3% | 47.5% | 55.5% | 18% | 28% | 52% |
| Worcester | 44,640 | 26,811 | 55.3% | 60.1% | 63.7% | 57.6% | 62.4% | 17% | 19% | 62% |
Mail, early and election day are each county's central forecast split by how ballots are cast; provisional ballots make up the rest.
How the model works
The model, fit in R, is a regression of ballots cast over active registration by county and party in six elections: the 2018 and 2022 midterms, the 2022 and 2026 primaries and the 2024 primary and general. It separates each election's statewide party level from a lasting pattern by county and party, with county census measures (education, age, mobility, home ownership). Each 2026 county-and-party group is then adjusted for how its registrants differ from 2022's in age and in how recently they registered, measured from the statewide voter list. The scenarios supply the statewide level.
Leaving one midterm out at a time and giving the model that year's true statewide party levels, its county turnout rates missed by these many percentage points on average. The naive forecast uses the other midterm's county pattern instead:
| Held out | Grain | Model | Naive |
|---|---|---|---|
| 2022 | county | 2.2 | 2.5 |
| 2022 | county and party | 2.7 | 3.0 |
| 2018 | county | 2.2 | 2.5 |
| 2018 | county and party | 2.6 | 3.0 |
What to keep in mind
- The spread between the low and high scenarios is the real uncertainty; the intervals within a scenario are narrow by construction.
- The central scenario's weights come from a written rule set before the numbers were in; the mail and early-ballot measures have no 2018 counterpart, and the early-ballot measure counts voters who moved from election day to early voting as more engagement.
- The early-voting count is the State Board's unofficial tally; mail ballots still to arrive are projected at 2022's rate of arrival.
- The 2026 primary had an incumbent governor, and its party split is estimated from the voter file.
- Composition is centred within party because the voter file lacks the 2022 voters who have since died or moved.
- Not modelled: election-day weather, late changes in the races, changes to mail-voting rules, and registration after September other than same-day registration.
- The 2014 general election is left out.
- The State Board's 2018-2024 turnout sheets divide by active registration; its 2026 guide divides by all eligible voters. Both rates are shown.
Sources and data
The full technical report sets out the model, the
rule for the central scenario verbatim, every index and the checks the run passed. The code is in
scripts/turnout_model. Every output has a provenance file naming its sources and run date:
- backtest.csv (provenance)
- backtest_cells.csv (provenance)
- composition_factors.csv (provenance)
- early_vote_check.csv (provenance)
- election_levels.csv (provenance)
- forecast_county.csv (provenance)
- forecast_county_party.csv (provenance)
- forecast_statewide.csv (provenance)
- level_rule.csv (provenance)
- level_rule_text.csv (provenance)
- mode_split.csv (provenance)
- reference_levels.csv (provenance)
- same_day_registration.csv (provenance)
- structure_effects.csv (provenance)