Degree days are the standard measure of how much heating or cooling the weather demanded, and the single biggest swing factor in winter natural gas consumption. The Oct 2025 – Mar 2026 heating season recorded 3,359 US heating degree days against a 10-year normal of 3,446 — 2.5% warmer than normal. The EIA currently projects Oct 2026 – Mar 2027 at 3,392 degree days, 2.0% milder than normal. Figures are population-weighted US averages from the EIA, measured against a 65°F base.
Solid line is observed weather through July 2026; the dashed continuation is the EIA STEO forecast, not measured weather.
| Region | HDD | HDD normal | vs normal | CDD | CDD normal | vs normal |
|---|---|---|---|---|---|---|
| New England | 11.8 | 3.6 | +228% | 184.5 | 248.1 | -26% |
| Middle Atlantic | 3.0 | 1.3 | +131% | 283.9 | 309.2 | -8% |
| East North Central | 9.1 | 4.0 | +127% | 304.3 | 274.0 | +11% |
| West North Central | 3.4 | 10.1 | -66% | 373.8 | 314.3 | +19% |
| South Atlantic | 0.0 | 0.0 | — | 489.6 | 482.9 | +1% |
| East South Central | 0.0 | 0.0 | — | 457.7 | 447.3 | +2% |
| West South Central | 0.0 | 0.0 | — | 587.8 | 592.4 | -1% |
| Mountain | 4.7 | 10.2 | -54% | 423.7 | 419.8 | +1% |
| Pacific | 12.7 | 13.7 | -7% | 253.1 | 260.3 | -3% |
Population-weighted degree days for the nine US census divisions, from the EIA Short-Term Energy Outlook, against a 65°F base. "Normal" is the prior 10-year average for the same month. A single summer month shows very few heating degree days by definition — the seasonal totals above are the meaningful comparison. See how this feeds through to gas storage and heating costs.
A degree day counts how far a day's average temperature sits from 65°F, the temperature at which a typical building needs neither heating nor cooling. A day averaging 40°F contributes 25 heating degree days; a day averaging 85°F contributes 20 cooling degree days. Summed across a month or season and weighted by population, they compress the entire national weather picture into one number that tracks energy demand closely.
This is why gas traders watch degree-day forecasts more closely than temperature forecasts. Demand does not care that it was 30°F in Wyoming if nobody lives there — population weighting means a cold snap across the Midwest and Northeast moves the number far more than the same anomaly in the Mountain West. A winter running 10% above normal on heating degree days can add hundreds of billions of cubic feet to gas demand, which shows up first in storage withdrawals and then in price.
The 65°F base is a convention rather than a physical constant, and it is imperfect: better-insulated modern buildings need heat at lower outdoor temperatures than older ones, so the relationship between degree days and actual fuel use drifts slowly over time. The comparison against a 10-year normal, rather than a fixed historical average, partly absorbs that drift.
A heating degree day measures how far the average outdoor temperature falls below 65°F on a given day. If the day averages 45°F, that is 20 heating degree days. Add them up over a month or a season and you get a single number describing how much heating the weather demanded. Cooling degree days work the same way in reverse, counting degrees above 65°F.
Heating demand is the largest swing factor in winter natural gas consumption, and degree days are the cleanest measure of it. A winter running 10% colder than normal can add hundreds of billions of cubic feet to gas demand, drawing down storage and lifting prices — while a mild winter leaves storage full and prices soft. The same logic drives heating oil and propane. In summer, cooling degree days drive electricity demand and therefore gas burn for power generation.
The Oct 2025 – Mar 2026 heating season recorded 3,359 US heating degree days against a 10-year normal of 3,446 — 2.5% warmer than normal.
No. The EIA Short-Term Energy Outlook is a forecast product, so months beyond July 2026 are projections based on NOAA weather outlooks, not observed weather. They are drawn with a dashed line on the chart and labelled as forecast in the tables. Weather forecasts beyond a few weeks carry wide error bars — treat the winter-ahead figure as a planning assumption, not a prediction.