Leave the AC On All Day or Turn It Off? - 4 vs 8 Hours
Here's the short version.
If you'll be gone for a workday, turn it up or off — that saves real money.
If you're running to the store, leave it alone.
Everything in between is where the internet starts fighting with itself.
It's early August, the bill from July just landed, and I'd bet the number on it was worse than you expected. You're not imagining it. NEADA projected in June that the average U.S. household summer electric bill would hit nearly $800 this year — up 10.5% from last summer and roughly 40% higher than 2020.
So the question "should I leave the AC on all day?" isn't trivia anymore. Cooling is about 19% of a typical home's energy costs, per Consumer Reports.
Search it and you'll find confident answers pointing in opposite directions. Some say never shut it off. Some say always shut it off.
Here's the thing: they're not disagreeing about physics. They're describing different equipment and different lengths of time. Once you sort those two variables out, the contradiction mostly disappears.
Let's sort them out.
How long do you have to be gone before turning the AC off pays off?
For a central AC, the honest answer is: somewhere between 4 and 8 hours. Below that it's a wash. At 8 hours, the savings are consistent.
The best evidence comes from three building-systems researchers at the University of Colorado Boulder — Aisling Pigott, Jennifer Scheib, and Kyri Baker — who ran building-energy simulations and published the results through The Conversation (republished by PBS NewsHour in July 2023).
What they actually tested matters, so here it is:
| Setup | Detail |
|---|---|
| Home | 1,200 sq ft, typical insulation |
| Climates | Phoenix, AZ (hot/dry) and Atlanta, GA (hot/humid) |
| Equipment | single-stage central A/C, central air-source heat pump, minisplit heat pump |
| Strategy A | hold 76°F all day |
| Strategy B | 8-hour setback to 89°F |
| Strategy C | 4-hour setback to 89°F |
Their finding, in their words: "Even when the A/C temporarily spikes to recover from the higher indoor temperatures, the overall energy consumption in the setback cases is still less than when maintaining a constant temperature."
Annual savings with a conventional central A/C: up to 11%.
But the useful nuance is in the timing. The 8-hour workday setback delivered consistent savings across all three system types. The 4-hour setback was, in their words, "less straightforward."
That's as close to a break-even point as the U.S. literature gets: four hours is ambiguous, eight hours is a clear win.
Shorter absences have clearer answers:
| Time away | What the sources say |
|---|---|
| Under 15 minutes | No savings from shutting it off |
| Under 1 hour (window unit) | Leave it running |
| 2–3 hours or more (window unit) | Turn it off |
| 4 hours (central) | Ambiguous |
| 8 hours (central) | Turn it up or off — consistent savings |
One note on sourcing: the window-unit thresholds come from HVAC professionals quoted in consumer press, not from a public agency. I'll come back to why that matters later.
Wait — the federal page that answered this was deleted in July 2026
This one surprised me, so I checked it myself: as of August 4, 2026, energy.gov/energysaver/heat-and-cool/home-cooling-systems returns a 404. So does the DOE thermostats page.
If you've been reading AC advice online, you've absolutely seen the DOE numbers cited: start with an indoor temperature between 75–78°F during the day, and raise it 7°F when you're away. That guidance lived on the Department of Energy's Energy Saver pages.
Those pages are gone. Newsweek reported on July 3, 2026 that the Home Cooling Systems page had started returning "Page not found," with archived versions showing it live through late June and early July.
It wasn't one page, either. Canary Media reported on July 22, 2026 that more than 1,600 pages were removed from the Energy Saver section — DIY home energy assessments, appliance shopping guides, cool-roof installation guides. More than 300 of the deleted pages had drawn 160,000 page views in a 30-day span before they came down.
Izzy Pacenza of the Environmental Data and Governance Initiative told Canary Media: "I can't remember another time that, with DOE specifically, we've seen an entire domain go down."
Now, let me be careful about what this does and doesn't mean.
The removal followed New York City Mayor Zohran Mamdani urging residents to set ACs to 78°F during an extreme heat wave affecting 26 states. But Newsweek explicitly stated there is "currently no evidence that the Department of Energy removed the webpage because of Mamdani's comments." DOE didn't answer questions from either outlet.
And this is important: the deletion does not mean the advice was wrong or retracted. The 7–10°F setback figure and the 75–78°F band are still widely cited by utilities and consumer press. The page hosting them just isn't there anymore.
So if you see an article confidently linking to energy.gov for its thermostat advice, click the link. A lot of them are pointing at a 404.
Does your AC coast, or does it slam on and off?
This is the single variable that decides whether "leave it on" is good advice for you. And you can usually figure it out from your equipment's spec sheet or your utility rebate paperwork.
Skip the jargon for a second and think about driving.
- Single-stage systems are gas pedal to the floor or nothing. Full blast, then off, then full blast again.
- Two-stage systems have a cruising gear and a passing gear.
- Variable-speed systems (including minisplits and most modern heat pumps) are basically cruise control. They coast at low output for hours.
Here's how that plays out:
| Type | How it runs | Efficiency range | What setback does for you |
|---|---|---|---|
| Single-stage | 100% or off, short cycles, weakest moisture removal | ~13–15.1 SEER2 | Best case for setback — up to 11% annual savings in the CU Boulder study |
| Two-stage | High/low output, longer cycles, better moisture removal | ~15.2–17 SEER2 | Middle ground; longer low-output runs shrink the payoff |
| Variable-speed / minisplit | Modulates continuously, "cruises" at partial load | 17 up to 26 SEER2 | Smallest payoff from setback |
(The efficiency ranges come from manufacturer education pages — Lennox and Carrier — so treat them as classification, not as a savings claim. The setback column is from the CU Boulder research.)
And here's the counterintuitive part, straight from the researchers: the heat pump and minisplit systems "are more efficient overall but yield less savings from temperature setbacks."
Read that twice. Better equipment doesn't mean bigger setback savings. It means the baseline was already efficient, so there's less waste to claw back.
Savings also shrink "if the home is better insulated, the A/C is more efficient or the climate has less dramatic temperature swings."
So if you own a modern variable-speed system, the "just leave it on" crowd is closer to right for you. If you own a single-stage central unit — which is still extremely common — the setback crowd is right.
What about the startup surge?
You've heard this one: restarting the AC costs more than leaving it on, because the compressor draws a huge spike.
The spike is real. A single-stage compressor can pull five to seven times its normal running current at startup. But that surge lasts seconds, not hours. Set that against hours of compressor runtime you didn't pay for, and it isn't close.
You don't have to take that on faith, either. The CU Boulder simulation measured the recovery spike explicitly — and still found net savings.
Should you turn the AC off at night while you sleep?
Turning it fully off is usually the wrong move. Turning it up is the right one.
The measured evidence here is the twin-house experiment at the Canadian Centre for Housing Technology (CCHT) — two identical research houses side by side, one running a setback schedule and one holding a constant setpoint. This is a government-affiliated research facility, not a vendor test.
Manning et al. reported that in the summer 2003 cooling season, starting from a 72°F daytime setpoint and setting forward to 77°F at night and during work hours saved 11% seasonally, with 14% savings on the hottest day.
Note the schedule: nights and work hours. Nighttime setup was part of a strategy that measurably worked.
⚠️ One sourcing caveat: those numbers reach us through a 2016 ACEEE paper citing the 2007 study, since the original paper sits behind a paywall. Attribute it as "Manning et al. 2007, as cited by ACEEE."
For a target number, ENERGY STAR's sample schedule lists 82°F for sleeping (versus 78°F when home and 85°F when away). More on why that table needs an asterisk in a moment.
A few reasons not to just kill the system overnight:
- Humidity keeps climbing while the system is off — and in a humid climate that's a mold issue, not just a comfort issue.
- If heat wakes you up at 3 a.m., you'll run it at full blast for an hour to recover, which undoes the point.
- Safety comes before savings during a dangerous heat wave, especially with pets, older adults, or heat-sensitive household members.
So: raise the number, don't flip the switch.
Is holding 78°F cheaper than cycling 72°F on and off?
Holding a higher number wins, and it isn't especially close. Raising your setpoint is a bigger lever than fiddling with the on/off switch.
Do the math with the numbers we have:
- Raising the thermostat is worth roughly 3% per °F in summer (attributed to Consumer Reports and DOE in reporting).
- Going from 72°F to 78°F is six degrees.
- Meanwhile, the CCHT houses saved 11% seasonally with a five-degree setup at night and during work hours.
Now compare that to the fiddling strategy: turning your AC off for a 30-minute errand and back on saves you nothing at all, and under 15 minutes there's no payoff whatsoever.
One lever is worth double digits. The other is worth roughly zero.
The practical version: pick the highest number you're actually comfortable with and hold it, then set it higher when you leave. That beats any amount of switch-flipping.
About that 78°F number
ENERGY STAR's widely shared table lists 78°F when home, 82°F sleeping, 85°F when away. You've seen it quoted as "the government says 78."
Here's the caveat almost nobody reports: EPA clarified in 2019 that it was not recommending those temperatures — or any others — for the cooling season. The table was meant as "a starting point," to be adjusted "to fit your family's schedule and stay comfortable."
And the now-deleted DOE page framed it as 75–78°F, a band rather than a single number.
So there is no official federal decree that you must sit at 78. There's a sample schedule, a deleted range, and a lot of people quoting both as gospel.
How much does raising the thermostat 1 degree actually save?
About 3% per degree in summer. In dollars, reporting works it out like this: a household averaging $300 a month in summer energy costs saves roughly $10 a month per degree.
That's about two coffee runs for one degree you probably won't notice after a week.
Three degrees is $30 a month. Over a June-through-September cooling season, that's real money — and it's money you get without buying anything.
Two more numbers for scale:
- The canonical DOE figure: a 7–10°F setback for 8 hours a day is worth up to 10% a year on heating and cooling.
- The average U.S. residential electricity price is projected at 18.2 cents/kWh in 2026, up nearly 5% from 2025. For Q3 specifically, EIA puts it at about 18.51 cents versus 17.68 cents a year earlier.
⚠️ Don't stack the two savings figures. Eight hours of a 7–10°F setback is not the same thing as 24 hours of one degree. They're consistent with each other, but you can't add them together.
The cheapest upgrade in this whole article
Ceiling fans. DOE says a ceiling fan can make a room feel as much as 4°F cooler.
Research from the Florida Solar Energy Center found that using ceiling fans combined with raising the thermostat 2°F produced about 14% net savings in annual cooling energy use.
The catch, per Consumer Reports: fans "cool people, not rooms." A fan running in an empty room is pure waste. It only pays off if you raise the thermostat setting to go with it.
Why the answer flips in Florida: humidity and the 60% line
In a dry climate, letting the house get hot while you're away is fine. In a humid one, it's how you get mold.
Experts quoted by CBS News — from Stony Brook University, Arizona State University, and CU Boulder — draw exactly this line. In Arizona, letting the house warm up is no problem. In Florida, an extended shutdown risks mold growth, and raising the setpoint is preferable to turning the system off.
Here are the thresholds worth knowing:
- EPA recommends indoor relative humidity below 60%, ideally 30–50%.
- Above 60% RH, mold, dust mites, cockroaches, and bacteria all become significantly more viable indoors.
- When humidity exceeds about 70% at temperatures of roughly 70–90°F, mold can germinate on damp organic surfaces within 24–48 hours.
That last number is the one that changes behavior. A long weekend away with the system off, in August, in Georgia, is comfortably inside that window.
Worth noting: the CU Boulder simulation tested both Phoenix and Atlanta and still found setback savings in the humid case. The humidity concern is about mold and comfort, not primarily about energy. You're not paying more by setting back in Atlanta — you're taking on a moisture risk.
For vacations, HVAC contractors commonly recommend 80–85°F while away in summer, and no higher than about 80°F in very humid coastal areas, so the system still cycles often enough to pull moisture out. I couldn't find a public-agency source with a specific vacation setpoint, so treat that range as industry practice and the EPA's under 60% RH target as the actual anchor.
Do smart thermostats actually deliver the savings they promise?
They help — but the real-world number is smaller than the theoretical one.
ENERGY STAR certified smart thermostats save on average about 8% of heating and cooling bills, or roughly $50 per year.
To earn the label, a product's field data has to show a runtime reduction of at least 8% for heating and 10% for cooling, verified by EPA-recognized certification bodies using real installations across the country. Independent experts describe roughly 10% energy reduction versus a manual thermostat.
Compare that to the theoretical ceiling — 10% from DOE's setback guidance, 11% from the CU Boulder simulation — and you can see the gap.
Why the gap? ACEEE researchers have argued for over a decade that programmable thermostat savings are "only as good as the people using them." That's the actual title of a 2012 ACEEE paper.
There's a sharper finding in the ACEEE proceedings too: programmed thermostats produced 25% higher summer peak demand compared with the ENERGY STAR-specified cooling program, and consumers in hot-humid climates simply don't follow that program.
Translation: the hardware works. The schedules people actually leave on it are the weak link. A $50-a-year gadget doesn't beat you just deciding to hold 78 instead of 72.
An honest caveat: look at who wrote most of the results you'll find
I want to flag something about this topic specifically, because it explains why the advice out there is such a mess.
Search "should I leave my AC on all day" and most of page one is HVAC repair and installation companies. They sell service calls and equipment. That doesn't make them liars — the mechanical explanations in contractor blogs are often accurate — but it does mean they have a structural reason to discourage aggressive cycling of equipment, and most of their claims come with no cited methodology.
That's why this article leans on the CU Boulder simulation, the CCHT twin-house measurements, EPA/ENERGY STAR, EIA, and NEADA, and flags contractor-sourced items (window-unit thresholds, vacation setpoints, SEER2 ranges) where they appear.
A few other limits I'd rather state than hide:
① The CU Boulder study is a simulation, not a measurement. It's the best-designed public source on the exact question, but it's modeled.
② The CCHT twin-house numbers are measured — but they're Canadian, from 2003, and they reach us secondhand through a 2016 ACEEE citation.
③ The $800 summer bill figure is a forecast, revised upward mid-season. NEADA's April 2026 outlook projected a smaller increase before the June update raised it.
④ Nobody has published a break-even point for U.S. central AC as clean as "90 minutes." Four to eight hours is a range, not a rule, and it moves with your insulation, your climate, and your equipment.
Frequently asked questions
Q. Is it cheaper to leave the AC on all day or turn it off?
A. For a single-stage central AC and a workday-length absence, setting back or shutting off is cheaper — up to 11% in annual savings in the CU Boulder simulation. For absences under 15 minutes, for window units under an hour, and for variable-speed or minisplit systems, leaving it running is closer to right.
Q. How long should I be gone to turn the AC off?
A. Eight hours produces consistent savings across every system type tested. Four hours is ambiguous. Under 15 minutes, there's no savings at all. For a window unit, HVAC pros put the threshold at 2–3 hours.
Q. Should I turn my AC off at night?
A. Raise it instead of turning it off. Measured twin-house data showed 11% seasonal savings from a 72°F to 77°F setup at night and during work hours. ENERGY STAR's sample schedule lists 82°F for sleeping.
Q. Is it cheaper to keep the AC at 78 or turn it on and off?
A. Holding a higher setpoint wins. Every degree up is worth roughly 3% in summer, while short on/off cycles save little or nothing.
Q. How much does 1 degree save on the AC?
A. About 3% per °F. On a $300 monthly summer energy bill, that works out to roughly $10 a month per degree.
Q. What should I set my thermostat to when on vacation?
A. EPA's defensible target is indoor humidity under 60%. HVAC contractors commonly suggest 80–85°F while away, and no higher than about 80°F in humid coastal areas so the system still runs enough to dehumidify. There's no public-agency vacation setpoint.
Q. Why was the energy.gov cooling page deleted?
A. DOE hasn't said. Newsweek reported the removal on July 3, 2026, and Canary Media reported on July 22 that over 1,600 Energy Saver pages came down. DOE did not respond to questions from either outlet. The pages return 404 as of August 4, 2026.
Q. Does a dirty filter really matter?
A. Yes. Change or clean the AC filter monthly during the cooling season, per Cornell Cooperative Extension. Consumer Reports says every 30 to 90 days, checked monthly during peak summer heat.
One thing to change before this weekend
Don't reorganize your whole cooling strategy. Do one thing.
Raise your setpoint one degree and leave it there.
One degree is about 3% — roughly $10 a month on a $300 bill, and you'll stop noticing it in a few days. That's a bigger, more reliable win than any amount of turning the unit on and off as you come and go.
If you want a second thing: turn on the ceiling fan in whatever room you're actually sitting in, and turn it off in the rooms you're not. Fans cool people, not rooms. Paired with a 2°F increase, Florida research measured about 14% in annual cooling savings.
And if you're leaving for a full workday tomorrow, push the setpoint up before you walk out. That's the eight-hour case, and it's the one the research is clearest about.
Which camp are you in?
① Leave it running, always ❄️
② Off the second I walk out the door 🔌
③ Trying the "one degree up" thing this week 🌡️
Drop a 1, 2, or 3 in the comments — and tell me what your July bill looked like. I suspect we're all in the same boat this summer 😊
Comments
Post a Comment