Which Connecticut county loses power the most?
Litchfield County. In 2024, the average thousand households there absorbed 7,265 customer-hours of outage, more than five times New Haven County's 1,337. And in 2023 the answer was different: Windham County, at 15,047 per thousand households, a number one December storm mostly built by itself.
If you expected Fairfield County and its coastline to top this list, that is the point of this post. Counted per household, Connecticut's outage burden lives in the quiet counties.
The 2024 ranking, per 1,000 households
A customer-hour is one metered account, roughly a household, without power for one hour. We divide each county's total by its households so small counties and big ones can be compared fairly. These come from federal outage records (DOE EAGLE-I) divided by Census ACS household counts; the full dataset is on the open data page.
| Rank | County | Customer-hours per 1,000 households | Share from its worst 3 days |
|---|---|---|---|
| 1 | Litchfield | 7,265 | 52.2% |
| 2 | Windham | 5,499 | 34.1% |
| 3 | Tolland | 4,938 | 23.6% |
| 4 | Middlesex | 3,848 | 36.2% |
| 5 | Fairfield | 3,075 | 29.8% |
| 6 | Hartford | 2,619 | 39.8% |
| 7 | New London | 1,958 | 7.2% |
| 8 | New Haven | 1,337 | 14.3% |
Source: DOE EAGLE-I raw 15-minute outage intervals, 2024 (all days, including the major storm days that filed reliability statistics set aside), divided by ACS 2021 household counts (county households range from 45,425 in Windham to 356,529 in Hartford). Computed 2026-07-15 under methodology v1.4. Every number in this table is machine-checked against the dataset on each site build.
The pattern that holds in both years
The three most urban counties, New Haven, Hartford, Fairfield, sit in the bottom half both years. The rural northeast and northwest corners carry the burden: Windham was first in 2023 (15,047) and second in 2024; Litchfield was third in 2023 (5,817) and first in 2024. New Haven County was the least-burdened both years (1,414 in 2023, then 1,337).
Why would rural counties lose more power per household? The standard industry explanation is longer lines through more trees serving fewer customers, so each failure takes longer to find and fix and each household shares more exposed wire. Our data does not test that explanation, so treat it as the common account, not our finding. What our data does show, without needing a theory, is the size and the consistency of the gap.
One storm decided first place, both years
Look at the last column. Litchfield took first in 2024 with 52.2 percent of its whole year concentrated in its worst three days. Windham took first in 2023 with 77.9 percent in the December 18 to 20 storm. The crown moves because the storms move, and the effect reaches below the crown too: New London's 2023 rate was itself mostly storm-built and halved in 2024, dropping it two places. That is the shape of grid risk in Connecticut: the day-to-day grid is genuinely good, and the year is defined by whether a big one hit your county.
What to do with this
If you live in the top half of that table, especially with a well pump, a sump pump, or anything medical in the house, your county's storm tail is a planning fact worth an hour of your time. Your town's page carries the record and its limits. Storm-driven risk alerts, scored publicly for accuracy, are coming; the waitlist opens on this site first.
Where these numbers come from
Every statistic above is declared in this post's source file and machine-checked against the open dataset on every build; if the data changes, the build fails until the post is corrected. Customer-hours: DOE EAGLE-I raw 15-minute intervals. Households: ACS 5-year 2021. Definitions, floors, and limitations: how we compute things.