Last reviewed: September 2026 · By the HVAC Brief Editorial Team · Our sourcing and editorial standards
The median American household sets the air conditioning to 72°F when someone is home, and only 12.8% set 78°F or warmer. The most common single setting is 70°F, in 16.8% of air-conditioned homes.
The 78°F figure that dominates this argument is not a Department of Energy comfort recommendation. DOE’s Energy Saver guidance publishes a winter number and no summer one. Almost nine in ten American homes are cooler than 78°F.
Thermostat settings are argued about constantly and measured rarely. The argument usually cites 78°F as though it were an official recommendation, and the reply is usually one person’s own thermostat. Neither side has the distribution.
The distribution exists. EIA’s Residential Energy Consumption Survey asks every sampled household what it sets the thermostat to, in summer and winter, for three different conditions: someone home during the day, nobody home during the day, and at night. The 2020 cycle covers 16,171 households answering the summer question, weighted to represent 109,511,484 homes that use air conditioning. It has not been published as a distribution before.
What Americans set in summer
The mean summer setting with someone home is 72.1°F, with a standard error of 0.04°F. The median is 72°F. The distribution is not a smooth curve: households cluster hard on round and even numbers, with 16.8% at 70°F and 15.8% at 72°F.
Against that distribution, the 78°F argument resolves quickly. 12.8% of air-conditioned households set 78°F or warmer when someone is home, 7.3% set exactly 78°F, and 4.6% set 80°F or warmer. 55.5% set 72°F or cooler.
Where 78°F comes from, and what DOE actually says
DOE’s Energy Saver guidance on thermostats states that you can “save energy in the winter by setting the thermostat to around 68°F to 70°F while you’re awake and setting it lower while you’re asleep or away from home.” It gives no equivalent summer figure. The 78°F number in wide circulation derives from a cooling setpoint used as a modelling default in energy code work, not from consumer guidance, a point covered in our explainer on what the federal thermostat guidance actually says.
That distinction matters for how the number should be used. A modelling default describes an assumption inside a simulation. It was never a finding about what households do, and the data show households do something else.
Winter, where DOE does publish a number
Winter is the more interesting test, because there is a federal figure to compare against. The mean winter setting with someone home is 70.0°F and the median is 70°F, from 17,745 households weighted to 117,738,794 homes.
46.8% of households sit inside DOE’s 68°F to 70°F band. 36.9% set warmer than 70°F, and 34.3% set 68°F or cooler. So the single most-cited piece of federal thermostat guidance describes slightly under half of American homes, and the largest deviation is in the direction of more heating, not less.
| Condition | Mean | Median | Standard error |
|---|---|---|---|
| Summer, someone home during the day | 72.1°F | 72°F | 0.04 |
| Summer, no one home during the day | 73.6°F | 74°F | 0.05 |
| Summer, at night | 71.4°F | 72°F | 0.04 |
| Winter, someone home during the day | 70.0°F | 70°F | 0.04 |
| Winter, no one home during the day | 67.7°F | 68°F | 0.05 |
| Winter, at night | 68.3°F | 68°F | 0.04 |
One row in that table is counter-intuitive and worth stating plainly. The summer night setting, 71.4°F, is cooler than the summer daytime setting when someone is home, 72.1°F. Americans cool their houses more for sleeping than for waking.
The setback almost nobody makes
Setback is the entire mechanism behind DOE’s advice to set it “lower while you’re asleep or away from home,” and it is the main behavioural lever a programmable thermostat exists to pull. The data show it is mostly unused.
The average summer setback, measured as the away setting minus the home setting, is 1.49°F. 61.7% of air-conditioned households report no summer setback at all, meaning the away setting is the same as or cooler than the home setting. Only 15.6% set back 5°F or more.
Winter is better but not by much. The average winter setback is 2.30°F, 52.7% of households report none, and 21.9% set back 5°F or more.
Thermostat hardware barely moves the outcome
If setback were a hardware problem, smart thermostats would show it. They show a difference, and it is small.
| Thermostat type | Households | Mean summer setting | Mean summer setback |
|---|---|---|---|
| Programmable | 48,868,096 | 72.4°F | 1.34°F |
| Manual / non-programmable | 38,558,408 | 71.6°F | 1.20°F |
| Smart or internet-connected | 12,257,498 | 72.9°F | 2.41°F |
| No thermostat | 9,827,481 | 71.8°F | 2.23°F |
Homes with a smart or internet-connected thermostat set back 2.41°F against 1.20°F for manual thermostats. That is double, and it is still only 2.4°F. Programmable thermostats, in 48,868,096 homes, set back 1.34°F, barely distinguishable from a manual dial.
Regional spread is larger than any other split
Climate explains more of the variation than income, hardware or anything else measured here.
| Census division | Mean setting | Share at 78°F or warmer | Sampled households |
|---|---|---|---|
| Mountain South | 74.9°F | 35.1% | 831 |
| Pacific | 73.7°F | 27.1% | 1,509 |
| South Atlantic | 72.9°F | 12.8% | 3,065 |
| West South Central | 72.5°F | 10.1% | 1,735 |
| West North Central | 71.8°F | 9.3% | 1,730 |
| Mountain North | 71.7°F | 11.4% | 943 |
| East South Central | 71.4°F | 6.6% | 1,265 |
| East North Central | 71.4°F | 8.4% | 1,889 |
| Middle Atlantic | 70.4°F | 8.6% | 1,806 |
| New England | 70.2°F | 8.5% | 1,398 |
The gap between the Mountain South division at 74.9°F and New England at 70.2°F is 4.7°F. On the 78°F threshold the spread is wider still: 35.1% of Mountain South homes against 6.6% in the East South Central division.
| Climate zone | Mean setting | Sampled households |
|---|---|---|
| Hot-Dry | 74.6°F | 1,404 |
| Hot-Humid | 73.6°F | 2,318 |
| Mixed-Dry | 72.1°F | 130 |
| Marine | 72.1°F | 512 |
| Mixed-Humid | 71.5°F | 5,263 |
| Cold | 71.1°F | 6,250 |
| Very-Cold | 70.1°F | 288 |
The Subarctic zone is excluded from that table. Six sampled households fall in it, below EIA’s threshold of 10 for publication.
Income runs the other way from the assumption
| Annual household income | Mean summer setting | Share at 78°F or warmer |
|---|---|---|
| Under $25,000 | 71.4°F | 13.9% |
| $25,000 to $49,999 | 72.1°F | 13.8% |
| $50,000 to $99,999 | 72.3°F | 12.4% |
| $100,000 to $149,999 | 72.4°F | 12.4% |
| $150,000 or more | 72.6°F | 11.3% |
Mean setting rises gently with income, from 71.4°F under $25,000 to 72.6°F at $150,000 or more. The share of homes running at 78°F or warmer moves the opposite way, from 13.9% to 11.3%. The lowest-income households are the most likely to be running hot, and the difference is small enough that it should not be over-read.
Frequently asked questions
What temperature do most Americans set their thermostat to in summer?
The median summer setting when someone is home is 72°F and the mean is 72.1°F, from EIA’s 2020 Residential Energy Consumption Survey. The single most common setting is 70°F, reported by 16.8% of households that use air conditioning, followed by 72°F at 15.8%.
Do most people really set their thermostat to 78 degrees?
No. 12.8% of US households that use air conditioning set 78°F or warmer when someone is home during the day, and 7.3% set exactly 78°F. 87.2% set cooler than 78°F. 55.5% set 72°F or cooler.
Does the Department of Energy recommend 78 degrees?
Not as a comfort recommendation. DOE’s Energy Saver guidance gives a winter figure, “around 68°F to 70°F while you’re awake and setting it lower while you’re asleep or away from home,” and publishes no equivalent summer setting. The familiar 78°F traces to a cooling default used in energy code modelling rather than to DOE consumer guidance.
What do Americans set the thermostat to in winter?
The median winter setting when someone is home is 70°F and the mean is 70.0°F. 46.8% of households sit inside DOE’s stated 68°F to 70°F band, 36.9% set warmer than 70°F, and 16.2% set cooler than 68°F.
How many households actually turn the thermostat back when they leave?
Fewer than half. 61.7% of air-conditioned homes report the same or a cooler setting when nobody is home in summer, meaning no setback at all, and 52.7% report no setback in winter. The average summer setback is 1.5°F and the average winter setback is 2.3°F.
Do smart thermostats change how people set the temperature?
Only slightly. Homes with a smart or internet-connected thermostat set 72.9°F in summer against 71.6°F for homes with a manual thermostat, and they set back 2.4°F when away against 1.2°F for manual. Smart thermostats were in 10.5% of households in 2020.
Which part of the country runs the warmest thermostat?
The Mountain South division, covering Arizona, New Mexico, Nevada and Utah, averages 74.9°F in summer with 35.1% of homes at 78°F or warmer. The East South Central division averages 71.4°F with 6.6% at 78°F or warmer.
Do higher-income households keep their homes cooler?
Slightly warmer on average, but they are less likely to run hot. Households earning $150,000 or more average 72.6°F in summer with 11.3% at 78°F or above, against 71.4°F and 13.9% for households earning under $25,000.
How reliable are these numbers?
They come from a federal probability sample of 18,496 households weighted to represent 123.5 million US homes. Standard errors are computed with EIA’s documented Jackknife method using the survey’s 60 replicate weights; the national summer mean has a standard error of 0.04°F. The data were collected in 2020 and are the most recent RECS cycle published.
Is this the same as the temperature people are comfortable at?
No. RECS records the thermostat setting or the reported indoor temperature, not a comfort preference and not the temperature the house actually holds. A setting is what the equipment is asked to do; whether it achieves it depends on the system, the ducts and the envelope.
Methodology and limitations
- Source: EIA Residential Energy Consumption Survey 2020, public use microdata file v7, released January 2024. 18,496 sampled households weighted to 123,529,025 occupied US housing units.
- Variables: TEMPHOMEAC, TEMPGONEAC, TEMPNITEAC for summer and TEMPHOME, TEMPGONE, TEMPNITE for winter. EIA defines each as the “thermostat setting or temperature in home” for the stated condition. Valid responses run 50 to 90°F; the code −2, Not Applicable, is excluded, which removes households not asked the question.
- Weighting: all estimates use the final analysis weight NWEIGHT. Unweighted counts are reported alongside so readers can see the sample behind each figure.
- Standard errors: computed with the Jackknife method EIA documents for this file, using the 60 replicate weights NWEIGHT1 to NWEIGHT60, with variance ((R−1)/R) times the sum of squared deviations of the replicate estimates from the full-sample estimate, R = 60, 59 degrees of freedom.
- Suppression: EIA advises against publishing estimates with a relative standard error above 50 or fewer than 10 sampled households. Both rules are applied here, which is why the Subarctic climate zone is omitted.
- Imputation: EIA imputed Don’t Know and Refuse responses and recommends using the imputed values to avoid biased estimation. The imputed values are used.
- No state-level figures are published here. EIA states that the RECS sample was not designed to estimate every variable at state level and that some state estimates may be unreliable on sample size. Census divisions and climate zones are the finest geography used.
- Setback is computed as a difference between two reported settings within the same household, not as an observed schedule. A household reporting the same away and home setting is treated as making no setback.
- Vintage: RECS 2020 data were collected in 2020 and this is the most recent published cycle. Settings may have shifted since, particularly given electricity price movements after 2021. The next cycle will supersede these figures.
- A thermostat setting is not an indoor temperature. RECS records what the thermostat is set to or what the respondent reports the indoor temperature to be, which is not necessarily what the house holds.
Sources
- US Energy Information Administration, Residential Energy Consumption Survey 2020 public use microdata, file recs2020_public_v7.csv, and the accompanying variable and response codebook v7.
- US Energy Information Administration, Using the Microdata File to Compute Estimates and Standard Errors, June 2023, for the Jackknife variance formula and the publication standards applied.
- US Department of Energy, Energy Saver, Programmable Thermostats, archived copy, for the 68°F to 70°F winter guidance.
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