Last reviewed: September 2026 · By the HVAC Brief Editorial Team · Our sourcing and editorial standards
Summer 2026 ran 25% hotter than normal across the United States, measured in population-weighted cooling degree days. By 26 September the year had already passed the full-year total of 31 of the last 41 years.
The largest anomalies hit cool-climate states, led by Colorado at +145%. Those states also have far more homes without air conditioning, though across all states the link between heat and missing cooling is weak.
NOAA’s Climate Prediction Center accumulates cooling degree days over the calendar year, from 1 January. Unlike the heating season, which only began on 1 July, the 2026 cooling season is substantially complete, so the question of whether this summer was unusually hot can be answered now.
How hot 2026 was
Through 26 September the United States accumulated 1,412 population-weighted cooling degree days, +284 above normal, or +25%. Against the same point in 2025 it was +144, or +11%.
The dashed line is 2026 with three months still to accumulate. It already sits above 31 of the 41 completed years in the record. The hottest full year on record here is 2016 at 1,525; the coolest is 1992 at 1,059.
The long-run trend runs opposite to heating
Cooling demand has risen. The first ten years of the record averaged 1,262 cooling degree days and the last ten 1,403, up 11.2%, with a least-squares trend of +39 per decade. Our heating degree day index shows heating demand falling over the same span. The installed base is being asked to do more cooling and less heating than it was sized for a generation ago.
Where it was hottest against normal
| State | Cooling degree days | Versus normal | Percent | Homes without AC |
|---|---|---|---|---|
| Colorado | 668 | +395 | +145% | 18.3% |
| Wyoming | 473 | +177 | +60% | 29.3% |
| Oregon | 345 | +110 | +47% | 23.7% |
| Maine | 330 | +102 | +45% | 29.6% |
| Washington | 287 | +89 | +45% | 47.5% |
| Montana | 354 | +106 | +43% | 34.9% |
| Virginia | 1,435 | +417 | +41% | 5.4% |
| West Virginia | 1,059 | +306 | +41% | 10.9% |
| Minnesota | 665 | +184 | +38% | 5.7% |
| Utah | 916 | +253 | +38% | 8.1% |
| New Jersey | 1,020 | +266 | +35% | 3.5% |
| North Carolina | 1,788 | +461 | +35% | 9.4% |
The top of that table is dominated by cool-climate states: Colorado, Wyoming, Oregon, Maine, Washington, Montana. Their normal cooling loads are small, so an unusually warm summer moves them by large percentages. Only Vermont among the contiguous states finished below normal.
| State | Cooling degree days | Versus normal | Homes without AC |
|---|---|---|---|
| Hawaii | 3,505 | +11% | 43.0% |
| Texas | 3,039 | +27% | 5.0% |
| Florida | 2,984 | +7% | 4.2% |
| Arizona | 2,731 | +4% | 5.8% |
| Louisiana | 2,536 | +9% | 7.4% |
| Mississippi | 2,501 | +29% | 7.1% |
| Oklahoma | 2,423 | +35% | – |
| Arkansas | 2,250 | +34% | 6.6% |
| South Carolina | 2,178 | +29% | 4.9% |
| Alabama | 2,151 | +23% | 6.5% |
Did the heat land where air conditioning is missing?
This is the question the data invites, and the answer needs care.
The six largest anomalies all fell in states where between 18% and 48% of homes have no air conditioning, according to EIA’s housing survey. Washington, at 45% above normal, has 47.5% of homes without cooling. By contrast, Texas, Florida and Arizona, where only 4% to 6% of homes lack it, ran much closer to their normals in percentage terms.
Across all 44 contiguous states with publishable figures, however, the relationship is weak. The correlation between a state’s anomaly and its share of homes without air conditioning is +0.16, or +0.26 on ranks. The third of states with the least air conditioning averaged +37% against normal; the best-equipped third averaged +26%.
So the accurate statement is narrow: the most extreme heat of 2026 fell disproportionately in poorly air-conditioned states, but this was not a general pattern across the country. State air conditioning shares are in our analysis of air conditioning and heat pumps by state, and households that could not use their cooling at all are in our energy insecurity report.
Frequently asked questions
Was summer 2026 hotter than normal?
Yes, by a wide margin. Through 26 September 2026 the US accumulated 1,412 population-weighted cooling degree days, +284 above the NOAA normal, or +25%. It was also +11% above the same point in 2025.
How does 2026 compare with previous years?
With three months of the year still to come, 2026 had already accumulated more cooling degree days than the full-year total of 31 of the 41 years from 1985 to 2025.
What is a cooling degree day?
A cooling degree day measures how far the daily mean temperature rises above 65°F. A day averaging 80°F contributes 15 cooling degree days. Summed over a period, it measures how much cooling the weather demanded, which is why utilities use it to forecast air conditioning load.
Is US cooling demand rising?
Yes. The mean of the first ten years in this record was 1,262 cooling degree days; the mean of the last ten is 1,403, up 11.2%. A least-squares trend across 41 years runs at +39 cooling degree days per decade. Our heating degree day index shows heating demand moving the other way.
Where was it hottest relative to normal?
Colorado, at +145% above its normal, followed by Wyoming (+60%) and Oregon (+47%). These are cool-climate states with small normal cooling loads, so a modest absolute rise is a large percentage.
Did the heat hit states without air conditioning?
The largest anomalies did. All six states furthest above normal have between 18% and 48% of homes without air conditioning, against 4% to 6% in Texas, Florida and Arizona. Across all 44 contiguous states with publishable data, though, the relationship is weak: a correlation of +0.16. The third of states with the least air conditioning averaged +37% against normal and the best-equipped third +26%.
Was anywhere cooler than normal?
Almost nowhere. Among the contiguous states only Vermont (-19%) came in below normal.
Is the 2026 figure final?
No. Cooling degree days accumulate over the calendar year, so the 2026 total will rise further through December. Late-year cooling is small in most states, so the year-to-date figure is close to, and a floor on, the final total.
Methodology and limitations
- Source: NOAA Climate Prediction Center population-weighted cooling degree days. Full years use the December monthly file, which carries the 1 January to 31 December accumulation. 2026 uses CPC’s weekly state product through 26 September.
- A cooling degree day is the excess of the daily mean temperature above 65°F, summed over the period. Population weighting makes the national figure describe the cooling demand facing the average American.
- Cooling accumulates over the calendar year, whereas heating accumulates from 1 July. The two indexes are therefore not on the same seasonal basis.
- Comparing a year-to-date figure with completed years understates 2026, since cooling degree days will continue to accumulate through December. The comparison is a floor.
- The historical series is the CONUS state file; CPC’s weekly product used for 2026 also reports Alaska and Hawaii, which are excluded from the correlation analysis.
- Normals and the incalculable flag are CPC’s. CPC writes the flag as -9999 in some files and -999 in others; both are treated as not available.
- Homes without air conditioning come from EIA RECS 2020 microdata (AIRCOND = 0), tested against EIA’s reliability standard. Four states fail it and are excluded from the correlation.
- The correlation of +0.16 is Pearson’s r across 44 contiguous states; Spearman’s rank correlation is +0.26. Both are weak and neither implies causation. Excluding Colorado’s outlying anomaly gives +0.14.
- Air conditioning shares date from 2020 and the cooling data from 2026; penetration has likely risen since in the states where it was lowest.
Sources
- NOAA Climate Prediction Center, population-weighted cooling degree days, CONUS states.
- NOAA Climate Prediction Center, weekly degree day summaries.
- US Energy Information Administration, RECS 2020 public use microdata, variable AIRCOND.
The HVAC Brief is an independent trade publication. We do not sell equipment, take manufacturer advertising, or accept payment for placement in our research. Corrections welcome.
Leave a Reply