Carbon Intensity of Electricity: How Grid Emissions Are Measured
The carbon intensity of electricity is the carbon dioxide (CO2) released to produce each unit of electricity, usually written as pounds per megawatt-hour (lb/MWh) or grams per kilowatt-hour (g/kWh). For the U.S. grid in 2024, the U.S. Energy Information Administration (EIA) lists an emissions factor of 785 lb/MWh, which is 0.785 lb/kWh or about 356 g/kWh. It is a ratio: total CO2 from power plants divided by total electricity generated.
That simple ratio hides choices. Which plants count, which year, which region, and whether the electricity is measured at the plant or at your meter all change the answer. Two official sources can both be right and still differ. This article shows the calculation, compares the published figures, and explains what a grid-average number does and does not describe.
The calculation: CO2 divided by kilowatt-hours
Carbon intensity is an average over a period, so the arithmetic is the same at any scale:
carbon intensity = total CO2 emitted ÷ total electricity generated
EIA's electricity profile for the United States gives the two inputs for 2024: CO2 emissions of 1,537,073 thousand metric tons and net generation of 4,308,634,297 MWh. Dividing, 1,537,073,000 metric tons ÷ 4,308,634,297 MWh is about 0.357 metric tons per MWh, or about 357 g/kWh (about 786 lb/MWh). That is within 0.2% of the 785 lb/MWh the same profile lists (785 × 0.4536 ÷ 1,000 = 0.356 metric tons per MWh). The page does not say what causes the small gap.
The unit conversions help when you compare sources. One kilowatt-hour equals 3.6 MJ, so 356 g/kWh is about 99 g per MJ of electricity. See energy units: Btu, joule and kWh for the conversions. In the Comparator, grid electricity appears as a reference item with EIA's 2024 factor, and you can view it next to coal and natural gas.
Worked example: a 1 kW heater that runs for 3 hours uses 3 kWh. At 0.785 lb/kWh, that is 3 × 0.785 = about 2.4 lb of CO2 (1.1 kg) attributed to the grid.
As of November 13, 2026, EIA's profile page still shows 2024 data (released November 10, 2025) and lists the next release as November 2026, so a 2025 value was not yet available when this was written.
Two official sources, several numbers
EIA and the U.S. Environmental Protection Agency (EPA) both publish grid averages, built from different data. EIA draws on its own plant and utility survey forms. EPA's eGRID database, the Emissions & Generation Resource Integrated Database, comes from EPA's Clean Air Power Sector Programs data. The results are close but not identical.
| Source and scope | Data year | lb CO2 per MWh | g CO2 per kWh (converted) |
|---|---|---|---|
| EIA, United States Electricity Profile, emissions factor | 2024 | 785 | about 356 |
| EIA FAQ, U.S. net generation (about 0.81 lb/kWh) | 2023 | about 810 | about 367 |
| EPA eGRID, U.S. CO2 output emission rate | 2023 | 767.209 | about 348 |
| EPA eGRID, lowest subregion in the table | 2023 | 242.089 | about 110 |
| EPA eGRID, highest subregion in the table | 2023 | 1,543.073 | about 700 |
The first three rows sit within about 6% of each other (810 ÷ 767 = 1.06). The gap comes from different years and different databases, so quote a figure together with its source and year. The last two rows show something larger than a data gap: the national average masks a wide spread across regions.
Why the same kilowatt-hour can carry different emissions
eGRID splits the country into subregions, each a part of the grid that EPA says generally has similar emissions and resource mix characteristics. In the 2023 table, the highest subregion rate is about 6.4 times the lowest (1,543.073 ÷ 242.089 = 6.37). The U.S. figure of 767.209 lb/MWh is an average of very different places.
EPA defines Scope 2 emissions as indirect emissions associated with the purchase of electricity, steam, heat or cooling. Its guidance on reporting them shows how grid factors are used. Organizations are told to report two totals:
- Location-based. It applies the average emission factor of the grid that supplies the facility. In the United States, EPA recommends eGRID total output subregion factors, which average all plants feeding the grid, including baseload, intermediate and peaking units.
- Market-based. It reflects contracts the buyer has signed to procure power from specific sources, such as renewable energy certificates.
Worked example for the location-based method: a facility buys 10 MWh (10,000 kWh). In the lowest subregion, 10 × 242.089 = 2,421 lb of CO2 (about 1.1 metric tons). In the highest, 10 × 1,543.073 = 15,431 lb (about 7.0 metric tons). The electricity is identical, and the attributed emissions differ by a factor of 6.4.
What a grid-average number leaves out
Delivery losses. EPA's guidance says Scope 2 emission factors should not include transmission and distribution losses. EIA estimates those losses averaged about 5% of the electricity transmitted and distributed in the United States from 2018 through 2022. If 5% is lost, delivering 1 kWh takes about 1.05 kWh of generation (1 ÷ 0.95), so the emissions behind a delivered kilowatt-hour run about 5% above the generation-side figure: 0.785 × 1.053 = roughly 0.83 lb. That is an illustration that mixes years, not a published value.
Fuel supply chains. The factor counts emissions at the power plants. EPA's guidance says it excludes emissions from fuel extraction and other upstream activities. Those belong to a life-cycle view, covered in combustion versus life-cycle emissions, and the two numbers should never be added or swapped.
Biomass carbon. EPA's guidance states that eGRID grid-average factors include methane and nitrous oxide from burning biomass but not the biomass CO2, which is reported separately. A biomass plant's CO2 is therefore accounted for outside the grid-average factor. How biogenic carbon is treated more broadly is the subject of carbon-neutral fuel claims.
The power plants themselves. The factor is an average of every plant on the grid. How individual plants turn heat into electricity, and why their efficiencies differ, is explained in how power plants turn fuel into electricity.
Common mistakes
- Treating a national average as local. Subregion rates in the 2023 eGRID table range from 242.089 to 1,543.073 lb/MWh.
- Comparing numbers from different years or sources. EIA's 2024 and 2023 figures and EPA's 2023 rate are close but not interchangeable.
- Mixing up lb/kWh and lb/MWh. 0.785 lb/kWh and 785 lb/MWh are the same number. A factor of 1,000 is an easy slip.
- Confusing CO2 with CO2 equivalent. EIA's factor is for CO2. eGRID also lists methane, nitrous oxide and CO2 equivalent in separate columns. See greenhouse gases and fuel burning for how those gases compare.
- Adding upstream or life-cycle emissions to a combustion factor. The grid average counts the plants. Fuel supply is a different measure.
- Comparing grams per MJ of electricity with grams per MJ of fuel. The electricity figure divides plant CO2 by the electricity produced, so conversion losses are already inside it. A fuel's factor divides by the fuel energy burned.
- Reading an annual average as a live reading. The annual figure describes a year of generation, not the mix at any particular hour.
Frequently asked questions
What is the carbon intensity of the U.S. electric grid?
EIA lists 785 lb of CO2 per MWh for 2024, which is 0.785 lb/kWh or about 356 g/kWh. EPA's eGRID gives 767.209 lb/MWh for 2023 data. The two numbers come from different databases and years, so they are close but not identical.
How is grid carbon intensity calculated?
Divide the total CO2 emitted by power plants over a period by the total electricity they generated in that period. For 2024, EIA's profile lists 1,537,073 thousand metric tons of CO2 and 4,308,634,297 MWh of net generation, which works out to about 0.357 metric tons per MWh.
Why does carbon intensity vary by region?
Grids differ in the plants that supply them. eGRID subregions are groups of plants with similar emissions and resource mix characteristics, and their 2023 rates run from 242.089 to 1,543.073 lb/MWh. A national average hides that spread, so location-based accounting uses the regional factor where possible.
Does the grid factor include electricity lost in transmission?
No. EPA's guidance says Scope 2 emission factors should not include transmission and distribution losses. EIA estimates those losses at about 5% of the electricity transmitted and distributed over 2018 through 2022, so the emissions per delivered kilowatt-hour are somewhat higher than the plant-side figure.
Sources
- United States Electricity Profile 2024: U.S. Energy Information Administration. Accessed November 13, 2026.
- How much carbon dioxide is produced per kilowatthour of U.S. electricity generation?: U.S. Energy Information Administration. Accessed November 13, 2026.
- How much electricity is lost in electricity transmission and distribution in the United States?: U.S. Energy Information Administration. Accessed November 13, 2026.
- Emissions & Generation Resource Integrated Database (eGRID): U.S. Environmental Protection Agency. Accessed November 13, 2026.
- Summary Data, eGRID with 2023 Data: U.S. Environmental Protection Agency. Accessed November 13, 2026.
- Greenhouse Gas Inventory Guidance, Indirect Emissions from Purchased Electricity: U.S. EPA Center for Corporate Climate Leadership. Accessed November 13, 2026.
- Scope 1 and Scope 2 Inventory Guidance: U.S. EPA Center for Corporate Climate Leadership. Accessed November 13, 2026.
Spotted something wrong or out of date? Tell us through the corrections page. We answer within 10 business days.