Methodology

Average price: revenue divided by kilowatt-hours

EIA's average price is average revenue per kilowatt-hour: the money utilities and suppliers billed a group of customers, divided by the kilowatt-hours those customers used. It includes fixed monthly charges, fuel and purchased-power adjustments and riders, because all of them are billed revenue. It mostly excludes taxes collected for state and local governments, so a household's bill is usually somewhat higher. It is not a tariff rate: a home's marginal price for one more kWh can be higher or lower, and small users pay more per kWh than large users because the fixed charge is spread over fewer kWh.

We never average monthly prices. A 12-month or year-to-date price is the sum of the months' revenue divided by the sum of their sales, as EIA does.

Which period each figure covers

• Last 12 months (September 2025 to August 2026): our main measure for comparing states, because it covers every season once. Computed from EIA-861M monthly revenue and sales by state.
• Latest month (August 2026¹): the newest single month, identical to EIA's Electric Power Monthly Table 5.6.A. Months since January 2026 are preliminary until EIA's annual census replaces them.
• Annual: calendar years from EIA-861 (final; 2025 released October 7, 2026). Before 2010, annual figures are the sum of EIA's monthly survey; we checked them against EIA's API.
• Change: the last 12 months against the 12 months before; a single month against the same month a year earlier (never against the previous month, which mixes in the season).

Bills and usage

The average monthly bill is residential revenue divided by the number of residential customers and by 12; usage is kWh divided the same way. The 2025 state figures are EIA's own Average Monthly Bill table. EIA counts customer accounts (meters), not households, and its bill table counts accounts of rooftop-solar systems owned by third parties as customers too; in states with many such systems, such as California, that lowers the average bill and usage per account.

Our 12-month bill uses the same formula on the monthly data. Bills on utility pages are the utility's own revenue per customer.

Utilities and service parts

Utility figures come from EIA-861, the annual census of every distribution utility and supplier, with one record per utility per state. Each record is split into service parts that we never mix:

• Bundled (full service): the utility sells both the electricity and its delivery. The utility's average residential price on our pages is this part only. In retail-choice states it covers customers on default service.
• Delivery only: customers who buy their supply elsewhere pay the utility for delivery. We show it as delivery revenue per kWh, never next to bundled prices as if comparable.
• Energy only: what suppliers and community choice aggregators collected for supply.
• Texas retail providers: retail electric providers that bill energy and delivery together in the competitive part of Texas. Texas delivery utilities (Oncor, CenterPoint and others) report delivery only and have no residential price.

About 1,703 small utility records come from EIA's short form, which reports all customers together; we show their all-customer average, labelled as such, and never call it residential. The last 12 months for a utility come from EIA-861M, a monthly sample of large utilities: 152³ of the 177² utilities with pages are covered, and their 2025 sums match the annual census.

The counties a utility serves come from EIA-861 service-territory records: a county is listed if the utility reports customers there, even if it serves only part of it. EIA still uses Connecticut's eight former counties, so our Connecticut maps do too.

Retail choice

The share of residential customers with a supplier is the number of energy-only and community choice aggregator accounts (plus Texas retail provider accounts) divided by all residential accounts as EIA counts them. It includes opt-out municipal and community aggregation where it is reported as energy only. State commissions publish their own switching counts, with different dates and definitions; where we have them, the state guides show both.

Reliability, generation and other measures

• Outages: SAIDI (average outage minutes per customer in the year) and SAIFI (average number of interruptions), from EIA-861. State figures are EIA's published state totals, with and without major event days; utility figures are the utility's own report.
• Generation mix: net generation by power plants located in the state, by energy source (EIA-923 via EIA's state file). Power crosses state lines, so this describes what a state produces.
• What households report paying: the Census Bureau's American Community Survey asks households what they paid for electricity last month, in ranges. Census publishes no median; ours is interpolated within the range that holds the middle household and is labelled an estimate. Survey figures carry 90% margins of error.
• Where electricity goes in a home: EIA's Residential Energy Consumption Survey of 2020, the latest with state estimates. It was a pandemic year; the 2024 survey's consumption results are due in spring 2027.
• Weather: population-weighted heating and cooling degree days from NOAA's Climate Prediction Center, for the 48 contiguous states.
• Forecasts: EIA's Short-Term Energy Outlook gives national forecasts only. We always label them as forecasts, with the edition's date.

How we check the numbers

Each rebuild compares our results with figures EIA and the other agencies publish themselves: EIA's monthly tables for the latest month and year to date, its average bill table rebuilt from utility records, its API for the history before 2010, utility monthly sums against the annual census, and the Census, BLS, LIHEAP and state commission figures we use. A rebuild with a failed check is not published.

Corrections and updates

If you find an error, write to [email protected]. Corrections are listed with their date on the updates page, together with the release calendar.