Most homes in the United States and Canada start needing heat in October. That’s the month when normal temperatures first drop below the threshold used to calculate heating demand, according to thirty years of climate data. But “most” is doing real work in that sentence, because the actual window stretches across five full months depending on where you live.
The month the heating season opens, and why it is a month
Across the 63 U.S. states, Canadian provinces and territories documented here, heating demand opens in October in 29 of them, in September in 18, and in November in 7. Those numbers come from NOAA’s U.S. Climate Normals (1991-2020) and Environment and Climate Change Canada’s Canadian Climate Normals (1981-2010), and they’re not a guess about when people flip a switch. They’re a measurement of when the average temperature crosses a specific line.
That line is what climatologists call the base temperature. In the United States, NOAA uses 65 degrees Fahrenheit. In Canada, ECCC uses 18 degrees Celsius. The opening month, then, is simply the first month in a given location where the normal mean temperature falls below that base. Once it does, a building starts accumulating heating degree days, the running tally that stands in for how much a furnace or heat pump has to work.
Here’s the part that trips people up: this is a thirty-year average, not a forecast and not a calendar date. A “normal” mean temperature for October in a given county might be 58 degrees, comfortably under the 65-degree base. But that average is built from three decades of Octobers that included warm stretches and cold snaps alike. The first real cold night of any given year almost always arrives before the monthly average technically crosses the line. So when this page says a season “opens” in a month, it means the month, not a day you can circle on the calendar.
| Opening month | Number of territories |
|---|---|
| September | 18 |
| October | 29 |
| November | 7 |
The two extremes of that calendar sit far outside this table. Saskatchewan’s normal mean temperature drops below the Canadian base as early as August, while Arizona doesn’t cross its threshold until December. Everything else falls somewhere between those two bookends, clustered heavily around the September-to-November stretch that covers the bulk of North America’s population.
Where the season never opens, and where it never closes
Some places don’t fit the month-by-month pattern at all, because for them there’s no month where heating demand starts. It’s already there in every one of them. Six territories never close the season: Alaska, Alberta, Newfoundland and Labrador, Northwest Territories, Nunavut and Yukon. In each of these, no single month’s normal mean temperature ever climbs back above the base. There’s no “opening” month to name here, because the season doesn’t have an edge to open from. It’s a constant.
For a household in one of these territories, that reality shows up as a heating system that’s never fully idle, even during the warmest weeks of the year. Budgeting for heat isn’t a seasonal line item so much as a year-round baseline cost, with the only real variable being how much colder the coldest stretch gets on top of that floor.
Hawaii sits at the opposite extreme, and it’s worth being precise about what that means. The published heating degree day figure for Hawaii reads 0.0 in every single month of the year. That’s not a data gap or a rounding artifact. It’s a real, measured figure showing that the state’s normal temperatures never drop below the heating base at any point in the annual cycle. Practically speaking, Hawaii is the one place on this list with no heating season at all by this measure, not a short one or a mild one.
Put the two situations side by side and you get a useful reminder: the “opening month” concept only works where a location actually crosses the threshold and then crosses back. Where it never crosses back, or never crosses at all, the calendar model breaks down, and the household math changes accordingly.
How far apart the two ends really are
A heating degree day is a running total. Take the base temperature, subtract the day’s mean temperature, and if the result is positive, add it to the year’s tally. A day with a mean of 40 degrees against a 65-degree base contributes 25 heating degree days. Do that for every day of the year and you get an annual figure that works as a reasonable stand-in for fuel use. Roughly double the degree days for a similar house, and you’re looking at roughly double the heating fuel, all else being equal.
That’s why the spread between the mildest and coldest territories matters more than the calendar month alone. A place that opens its season in October but racks up a modest annual total is a very different heating proposition than a place that also opens in October but piles up degree days all winter long.
In the United States, using the 65 F base over the 1991-2020 normals period, Alaska logs 9,979 heating degree days a year while Arizona logs 874. Hawaii’s figure is 0. Excluding Hawaii from the comparison, since dividing by zero produces a meaningless ratio, the spread across the fifty states works out to roughly elevenfold between the coldest and mildest.
Canada’s numbers, calculated on an 18 C base over the 1981-2010 normals period, run from Nunavut’s 10,376 degree-days down to British Columbia’s 2,818, a spread of roughly four to one.
| Country | Base / Period | Coldest territory | Mildest territory | Spread |
|---|---|---|---|---|
| United States | 65 F, 1991-2020 | Alaska: 9,979 HDD | Arizona: 874 HDD (Hawaii: 0, excluded) | ~11 to 1 |
| Canada | 18 C, 1981-2010 | Nunavut: 10,376 HDD | British Columbia: 2,818 HDD | ~4 to 1 |
Notice these two rows never touch each other in a ratio. That’s deliberate. The American figures use a 65 F base measured over 1991-2020; the Canadian figures use an 18 C base measured over 1981-2010. Different scale, different reference point, different thirty-year window. What’s genuinely comparable between the two countries is the month the season opens. The degree-day totals themselves are not, and any site that lines up an American number against a Canadian one to declare one country “colder” is comparing two different rulers.
What the opening month is actually for
The opening month isn’t really a household deadline. It’s a maintenance deadline. The smart move is to treat that month as the window to test the heating system before it’s actually needed, not the moment to discover, on the season’s first genuinely cold night, that it doesn’t work. That particular night is also when every heating technician within fifty miles is already booked solid, sometimes for a week or more.
A practical pre-season sequence starts cheap and fast, and gets more involved only as needed. Here’s the order that matters, starting with what a homeowner can do in minutes and ending with when to call in a professional:
- Change or clean the air filter, since a clogged filter is the single most common cause of poor airflow and system strain.
- Clear physical clearance around vents, registers, and any outdoor unit components, removing furniture, curtains, leaves, or debris that’s accumulated over the warm months.
- Run a short test cycle, letting the system operate for ten to fifteen minutes while listening for grinding, banging, or rattling sounds that weren’t there before.
- Expect a faint burnt-dust smell on the very first firing of the season. That’s normal, caused by dust that settled on heating elements over the summer burning off, and it should fade within a few minutes.
- Test the carbon monoxide alarm by pressing its test button, and replace batteries if it’s battery-powered. Carbon monoxide is a real risk with combustion heating systems, and a working alarm is non-negotiable.
- If the burnt smell persists beyond a few minutes, if odd noises continue past the first cycle, or if the system fails to reach a comfortable temperature at all, stop and call a licensed technician rather than troubleshooting further.
Even a system in good working order is losing efficiency somewhere between the furnace and the vents. According to the U.S. Department of Energy, “in a typical house, about 20%-30% of the air that moves through the duct system is lost due to leaks, holes, and poorly connected ducts.” That’s a significant chunk of paid-for heated air never reaching the rooms it’s meant for, and it’s worth having ducts inspected as part of the same seasonal check, especially in a home that’s never had that done.
What turning it down is worth, in the Department of Energy’s own words
The Department of Energy puts a specific number on thermostat setbacks: “You can save as much as 10% a year on heating and cooling by simply turning your thermostat back 7-10 F for 8 hours a day from its normal setting.” That’s the exact statement, and it’s worth reading closely rather than paraphrasing.
The figure that circulates most widely online, that each degree of setback saves 7 to 10 percent, is a misreading of this statement. The Department’s number is an annual figure, it covers heating and cooling combined across the whole year, and it applies to a stated setback of 7 to 10 degrees held for 8 hours a day, not to each individual degree. Stretching that into a per-degree rule overstates the savings dramatically and misrepresents what was actually measured.
What does tie back to the opening month is duration. A household in a territory where the season opens in September and never closes, like the six territories with year-round heating demand, is running that thermostat setback for far more total hours across the year than a household where the season opens in November and wraps up early. The setback itself doesn’t change with the calendar. How many hours it gets to work does.