How Much Electricity Does a Heat Pump Use? kWh per Month and Year by Climate (2026)

How Much Electricity Does a Heat Pump Use?

By EnergyBS Editorial Team | November 13, 2026

A whole-house heat pump in a house that loses 500 Btu per hour for every degree of indoor-outdoor difference uses about 8,400 kWh a year in Chicago, 10,400 kWh in Minneapolis, 7,800 in Boston, 6,200 in Seattle and 3,600 in Atlanta, at a seasonal COP of 2.5. January is the heaviest month: 1,700 kWh in Chicago and 2,100 in Minneapolis. At EIA's 2026 regional residential prices, the Chicago figure is about $1,615 a year.

Those numbers come from three inputs you can check: NOAA's 1991-2020 heating degree day normals for each city, the house's heat loss per degree, and the heat pump's seasonal efficiency. Change any one and the answer moves in direct proportion. A tighter house that loses 300 Btu/h per degree uses 60% as much; a unit that averages COP 3.0 instead of 2.5 uses 83% as much. The tables below show the arithmetic, so you can swap in your own house rather than trust a national average.

Three Inputs, One Line of Arithmetic

Every kWh figure on this page comes from the same formula:

kWh = heat loss coefficient x 24 x heating degree days / 3,412 / seasonal COP

Each piece has a plain meaning.

Heat loss coefficient (Btu/h per degree F). How many Btu per hour the house leaks for each degree Fahrenheit the outside is colder than the inside. Engineers call it UA. A heating contractor's Manual J load report gives you the design heat loss, and dividing it by the design temperature difference gives UA. A house with a 35,000 Btu/h design load at 70F inside and 0F outside has a UA of 35,000 / 70 = 500. Our Manual J sizing guide shows how that load is calculated room by room.

Heating degree days. For each day, the number of degrees the mean outdoor temperature sits below 65F, summed over the month or year. A day averaging 25F adds 40 degree days. NOAA publishes 30-year normals for every major US weather station; Ontario's Building Code publishes Celsius degree days below 18C for Ontario towns. Multiply degree days by 24 hours and by UA and you have the heat the house needs, in Btu.

3,412. The number of Btu in one kWh, from EIA's conversion table. Dividing Btu of heat by 3,412 turns heat into kWh of heat.

Seasonal COP. The coefficient of performance: kWh of heat delivered for each kWh of electricity used. A COP of 2.5 means one kWh from the grid moves 2.5 kWh of heat indoors. The rated seasonal figure on US equipment is HSPF2, measured in Btu per watt-hour over a standard heating season, so COP = HSPF2 / 3.412. An HSPF2 of 7.8, the ENERGY STAR floor for split systems, works out to 2.29. An HSPF2 of 8.5, the ENERGY STAR cold-climate floor for non-ducted splits, works out to 2.49. HSPF2 10 is 2.93.

Put the constants together and you get a shortcut worth memorising: at COP 2.5, each 100 Btu/h per degree of heat loss uses about 281 kWh per 1,000 heating degree days. At COP 2.0 that becomes 352 kWh; at COP 3.0, 234 kWh.

So a house with UA 500 in a 6,000 degree-day climate uses 5 x 281 x 6 = 8,430 kWh a year at COP 2.5. That is the Chicago row below, within rounding.

Month by Month in Five US Climates

The table holds one house constant, UA 500 at a seasonal COP of 2.5, and changes only the weather. Degree days are NOAA's 1991-2020 monthly normals for each city's main airport station.

kWh of heat pump electricity for space heating only. UA 500 Btu/h per degree F, seasonal COP 2.5, base-65F degree days. Summer rows show the degree-day method assigning a little heat to cool nights; a real thermostat would call for almost none of it.

Three things in the table are worth knowing before you look at a bill.

December, January and February carry most of the year. In Minneapolis those three months hold 4,089 of 7,399 degree days, 55% of the total. Chicago and Boston come in at 56% and 53%. A heat pump bill that triples between October and January is behaving normally.

Seattle is spread out. Its January normal of 687 degree days is less than half of Minneapolis's 1,513, but its May still has 242 and its June 118, because Seattle's mean temperature stays below 65F well into summer (62F in June). The heat pump runs lightly for more months.

Atlanta's January is a bigger share of a small year. January is about 25% of Atlanta's annual degree days, against about 20% in the three northern cities, because Atlanta has little heating outside December to February.

Annual kWh by House and Efficiency

UA 500 is a mid-sized older house. The next table shows the range from a tight, smaller house (UA 300, about a 21,000 Btu/h design load at 70F inside and 0F outside) to a leaky or larger one (UA 750, about 52,500 Btu/h at the same conditions). Each row gives annual space heating kWh at three seasonal COPs.

The COP columns are not interchangeable across cities. A given heat pump runs at a higher average COP in Atlanta, where the January normal mean is 44.8F, than in Minneapolis, where it is 16.2F. For a cold-climate unit, read the COP 2.0 or 2.5 column in Minneapolis and Chicago, and the 2.5 or 3.0 column in Atlanta and Seattle. Boston sits between. That choice is a judgment, not a measurement, and the next sections show where it matters most. For the low-temperature COP of specific listed models, our cold-climate heat pump performance data collects the published figures.

The table also shows why insulation and air sealing move the kWh more than equipment upgrades usually do. In Chicago, going from UA 750 to UA 500 cuts 4,202 kWh a year at COP 2.5. Going from COP 2.5 to COP 3.0 on the UA 750 house cuts 2,101 kWh.

The Dollar Figure at 2026 Residential Prices

EIA's November 2026 Short-Term Energy Outlook puts the 2026 average residential electricity price at 18.17 cents per kWh nationally, and between 14.49 cents (West North Central, which includes Minnesota) and 28.84 cents (New England) by region. The table prices the COP 2.5 column from the previous section at each city's regional average. These are averages of whole bills divided by kWh, so they include delivery and fixed charges spread over every kWh.

Space heating only, seasonal COP 2.5. Cost = annual kWh x regional price. The heat pump's cooling and any resistance backup are extra.

The ranking is not the climate ranking. Boston has 25% fewer degree days than Minneapolis but the highest bill on the table, because New England's 28.84-cent average is twice the West North Central's 14.49 cents. Minneapolis heats the most and pays less than Boston at every house size.

The Seattle row is the one to adjust. EIA's Pacific region includes California, so the regional average can sit well above what a Washington household pays. A Seattle reader should multiply the 6,156 kWh by the per-kWh figure on their own bill; at half the Pacific average, the annual cost halves too.

The same caution applies everywhere. A regional average hides utility-to-utility differences larger than the gap between regions, and some utilities offer a lower winter rate or a heat pump rate. Our guide to reading an electric bill shows where to find the per-kWh charges that change with use, which is the number that belongs in this calculation.

For the comparison against keeping a gas furnace, including installed prices and a 15-year total, see our heat pump vs gas furnace true cost comparison. This page stops at the electricity.

Ontario and Quebec: Degree Days From the Building Code, Prices From the Regulator

Canadian degree days use a base of 18C, which is 64.4F, close enough to the US 65F base that multiplying Celsius degree days by 1.8 gives a usable Fahrenheit equivalent. Ontario's Building Code, Supplementary Standard SB-1, lists degree days below 18C for Ontario municipalities: 3,520 for Toronto City Hall, 3,800 for Etobicoke and Scarborough, 4,440 for Ottawa City Hall and 5,650 for Thunder Bay. SB-1 notes that its degree days were compiled for the 25 years ending in 2006, so they cover an older period than NOAA's 1991-2020 normals and may differ from recent winters.

Seasonal COP 2.5, space heating only. Toronto's suburbs at 3,800 degree days use about 8% more than the City Hall row.

Toronto sits between Chicago and Minneapolis on degree days, and Thunder Bay is colder than any US city in the first table. In Thunder Bay a seasonal COP of 2.5 is optimistic for most equipment; read the UA row and scale by 2.5 divided by your expected COP.

Ontario: the marginal price is the upper tier or the time-of-use period

The Ontario Energy Board's Regulated Price Plan prices from November 1, 2026 are 11.2 cents per kWh for the first 1,000 kWh a month in winter and 13.2 cents above that on the tiered plan, or 9.1, 14.6 and 18.9 cents for off-peak, mid-peak and on-peak on time-of-use. The ultra-low overnight plan charges 3.7 cents overnight and 37.7 cents on-peak. All of these are commodity prices; your distributor's delivery charge, part of which varies with kWh, comes on top.

A house that already uses a few hundred kWh a month on lights, appliances and hot water will push most of its winter heat pump use above the 1,000 kWh threshold, so the 13.2-cent upper tier is the marginal rate for the tiered plan. On that basis the commodity cost for the UA 500 Toronto house is 8,913 x 0.132 = $1,177 a year (CAD), and the Ottawa house is $1,484, before delivery charges and the Ontario Electricity Rebate. On time-of-use, the cost depends on when the compressor runs. A heat pump holds a steady setpoint around the clock, and nights and weekends are off-peak, so a large share of its use falls at 9.1 cents. Our Ontario electricity rates explainer walks through choosing between the plans, and the Ontario heat pump cost guide covers installed prices and rebates.

Quebec: Rate D's 40 kWh per day decides which price the heat pump pays

Hydro-Quebec's Rate D, effective April 1, 2026, charges a system access charge of 46.154 cents a day, 7.065 cents per kWh for the first 40 kWh multiplied by the days in the billing period, and 11.142 cents per kWh for the rest. The threshold is applied to the whole period, so a 62-day bill has a 2,480 kWh first tier.

A worked bill shows how this lands. Suppose a household uses 4,500 kWh over a 62-day winter period (an illustrative figure, not a typical one):

• First tier: 2,480 kWh x $0.07065 = $175.21
• Second tier: 2,020 kWh x $0.11142 = $225.07
• Access charge: 62 days x $0.46154 = $28.62
• Total before taxes: $428.90

Every additional kWh the heat pump uses in that period costs 11.142 cents, because the first tier is already full. In an electrically heated home whose winter use already passes 40 kWh a day, the second-tier price is the right one for heating estimates. At that price, a house with Ottawa's degree days and UA 500 would pay 11,243 x 0.11142 = $1,253 a year for heat pump electricity. Montreal's own degree days are not in Ontario's table, so a Quebec reader should use the bill method in the next section, which needs no degree-day figure at all.

Quebec households on Rate Flex D pay 46.463 cents per kWh during winter peak events and 4.886 cents (first 40 kWh a day) or 9.103 cents otherwise. A heat pump that keeps running through a peak event costs about four times Rate D's 11.142-cent second-tier price for those hours, so the plan pays only if the house can coast through events with the heat pump turned down.

Turning Last Winter's Fuel Bill Into Heat Pump kWh

Degree days and an assumed UA give a starting point. Your own heating history gives a better one, because it already includes your house's air leaks, your thermostat habits and your local weather. The conversion has five steps:

1. Total the heating fuel for 12 months from bills or delivery slips.
2. Remove the non-heating use. If gas also heats water, cooks or dries clothes, average the four lowest summer months and multiply by 12. That is the baseload; subtract it.
3. Convert to Btu of fuel. EIA's factors: 100,000 Btu per therm, 1,036 Btu per cubic foot of gas, 138,500 Btu per gallon of heating oil, 91,452 Btu per gallon of propane.
4. Multiply by the old system's efficiency to get the heat actually delivered to the house. The AFUE is printed on the furnace's rating plate; an 80% furnace delivers 0.80 of the fuel's Btu.
5. Divide by 3,412 and by the heat pump's seasonal COP.

Three worked cases follow. The household figures are illustrations with stated inputs; the conversion factors and degree days are sourced.

A Chicago gas house

The household used 950 therms over 12 months. June to September bills were 19, 17, 16 and 18 therms, an average of 17.5, so the baseload is 17.5 x 12 = 210 therms a year.

• Heating gas: 950 - 210 = 740 therms
• Fuel energy: 740 x 100,000 = 74.0 million Btu
• Delivered heat at 80% AFUE: 74.0 x 0.80 = 59.2 million Btu
• Heat pump at COP 2.5: 59,200,000 / 3,412 / 2.5 = 6,940 kWh a year
• At COP 2.0: 8,675 kWh
• At the East North Central 2026 price of 19.22 cents: $1,334 a year at COP 2.5

A Boston oil house

The house burned 750 gallons of heating oil in a year, and the boiler heats only the house (hot water is electric). Assume the boiler's rated efficiency is 80%.

• Fuel energy: 750 x 138,500 = 103.9 million Btu
• Delivered heat: 103.9 x 0.80 = 83.1 million Btu
• Heat pump at COP 2.5: 83,100,000 / 3,412 / 2.5 = 9,742 kWh a year
• At COP 2.0: 12,178 kWh
• At New England's 28.84 cents: $2,810 a year at COP 2.5, or $3,512 at COP 2.0

A propane house

A rural house used 600 gallons of propane for heat with a 90% furnace.

• Fuel energy: 600 x 91,452 = 54.9 million Btu
• Delivered heat: 54.9 x 0.90 = 49.4 million Btu
• Heat pump at COP 2.5: 5,789 kWh a year

The bonus: your own heat loss number

Once you have delivered heat, divide it by 24 and by the degree days for the same 12 months, and you have the house's UA. For the Chicago house, using the normal year: 59,200,000 / 24 / 5,974 = 413 Btu/h per degree F. For the Boston oil house: 83,100,000 / 24 / 5,546 = 624.

That number lets you read the monthly table directly. The Chicago house sits at 413 / 500 = 0.83 of the UA 500 column, so its January heat pump use is about 1,735 x 0.83 = 1,433 kWh at COP 2.5. The Boston house is 1.25 times the column, so its January is about 1,908 kWh. It also gives you a sanity check on a contractor's load calculation. A Manual J design load that implies a UA far above what your fuel history shows usually means an oversized system.

For the bill year to match the normal, use the degree days for the same 12 months the bills cover. Some gas utilities print the period's degree days on the bill; if last winter was 10% colder than normal, divide the heating fuel by 1.10 before converting. Canadian gas bills in cubic metres need the heat value your utility uses for that period, which most Canadian gas bills or utility rate pages state in gigajoules or megajoules per cubic metre; multiply cubic metres by that value, then convert 1 GJ to 277.8 kWh of fuel energy.

The steps assume the old system's efficiency held at its rating. An old furnace with leaky ducts in an unheated attic or crawlspace delivers less than its AFUE suggests, which makes the bill method overstate the heat the house truly needs. If the heat pump will use the same ducts, the loss carries over and the estimate stays fair. If you are switching to ductless heads inside the heated space, expect slightly less electricity than the calculation shows.

Where January Breaks the Seasonal Average

A seasonal COP is an average over a whole winter. The coldest weeks run below it, and that is where the kWh concentrate.

ENERGY STAR's cold-climate criteria for non-ducted split heat pumps require a COP of at least 1.75 at 5F. That is a floor for the category, not a typical value, but it shows how far efficiency can fall from a seasonal 2.5. Minneapolis's January normal mean temperature is 16.2F, and many January nights drop well below that.

Rerun January for the UA 500 Minneapolis house at different COPs:

The 912 kWh between the first and last rows is a fair estimate of what one hard January can add. It also explains a common surprise: the annual bill can match a forecast built on the seasonal COP while the January bill runs 25% to 43% above the forecast's January.

Two more effects show up in cold weather.

Resistance backup. When the heat pump cannot keep up with the load, electric strips or a backup furnace make up the gap. Electric strips run at a COP of 1, so each hour they carry the load uses 2.5 times the electricity the heat pump would have used for that same heat. A system sized close to the design load uses strips only on the coldest few days; one sized for cooling in a cold climate may use them for weeks. Our guide to heat pump backup heat and dual fuel covers lockout temperatures and how to keep strips from running when they are not needed.

Power draw at any moment. kW is the rate, kWh the amount. A heat pump delivering 36,000 Btu/h of heat at a COP of 2.5 draws about 36,000 / 3,412 / 2.5 = 4.2 kW. At COP 1.75 the same heat output takes 6.0 kW. In an hour at full output, that is 4.2 to 6.0 kWh. Most of the season it runs well below full output, because inverter compressors slow down when the load is small. If kW and kWh still blur together, our kW vs kWh explainer sorts them out.

Spotting the Heat Pump in Your Monthly Use

Once a heat pump is in, the bill shows total kWh, not heating kWh. Separating them takes one summer and one winter of data.

Take the average of the two or three months when the house neither heats nor cools, usually May and September or October in the northern US and Canada. That is the non-HVAC baseload: fridge, lights, cooking, laundry, hot water if electric. Subtract it from each winter month. What remains is mostly heat pump.

Then compare it with the monthly table. If the Chicago house above (UA 413) shows 1,450 kWh of heating in January against a predicted 1,433, the system is performing to plan. If it shows 2,200, something is pulling efficiency down: strips running too often, a dirty filter, a setback schedule that forces long full-speed recoveries, or a refrigerant problem. A one-month gap after an unusually cold spell is normal; a gap that persists through February and March is worth a service call.

Smart thermostats and many heat pump apps report runtime and, on some models, auxiliary heat runtime. If aux hours climb on days above your lockout temperature, that is where the extra kWh are going.

Why Your House May Not Match the Table

The degree-day method is simple on purpose. It has known biases, and knowing their direction helps you read the result.

The 65F base. Degree days assume the house needs heat whenever it is below 65F outside. A well-insulated house with many people and appliances inside may not need heat until it is 55F or colder outside, because internal heat covers the difference. For such a house the method overstates kWh, most in spring and fall. The bill method avoids this bias entirely, because it starts from what the house actually used.

Setpoint. The table assumes a roughly constant indoor temperature near 70F. Keeping the house at 66F instead lowers use most, in percentage terms, in mild months, when the indoor-outdoor gap is small, and least in January.

Ducts. A ducted heat pump with ducts in an unconditioned attic loses part of its output before it reaches the rooms, and the house has to make up the loss. That shows up as a higher effective UA.

Defrost. In humid weather near freezing, the outdoor coil ices up and the unit runs short defrost cycles that use energy without heating the house. The HSPF2 rating includes a defrost allowance for its test climate; a damp coastal winter can run more defrost than the test assumes.

Sizing. An oversized single-speed system cycles on and off often, and each start wastes some energy. Inverter systems avoid most of this but lose efficiency when they run at the bottom of their range for long periods. An undersized system leans on backup heat. Either way, actual kWh drift above the table.

Partial-house heating. A ductless unit that heats the main floor while a furnace or baseboard heats the rest uses only its share. Our mini-split running cost guide covers single-zone and multi-zone running cost by system size, which is the better reference when the heat pump serves part of the house.

Cooling and Hot Water Are Separate Loads

The heat pump also cools in summer, and that electricity is on top of the heating figures above. Cooling kWh follow a different rating, SEER2, and different weather, cooling degree days. In northern cities the cooling share is small next to heating; in Atlanta it can exceed the heating kWh. The mini-split guide linked above has a cooling table by system size and climate.

A heat pump water heater is a separate appliance with its own rating (UEF) and its own kWh. Do not mix its use into the space heating estimate.

A Worksheet for Your Own House

For line J, multiply UA by 24 and by your city's January degree days, then divide by 3,412 and by a January COP lower than the seasonal one. For the Chicago house: 413 x 24 x 1,233.5 = 12.2 million Btu; divided by 3,412 and by 2.0 gives 1,792 kWh, the planning figure for the year's heaviest bill.

If you are still deciding whether to switch, the next useful number is the federal or provincial money available for the switch, which changed in 2026; the heat pump rebate tracker by state shows which programs are still paying, and Canadian heat pump grants covers what is open in each province. Both sit with our other equipment, sizing and repair guides on the heat pumps hub.

Frequently Asked Questions

How many kWh does a heat pump use per day in winter?

Divide the month's kWh by its days. The UA 500 house at COP 2.5 uses about 69 kWh a day in a normal Minneapolis January (2,128 / 31), 56 in Chicago, 49 in Boston, 31 in Seattle and 28 in Atlanta. A day near 0F can run double the January average, because the heat loss is higher and the COP lower.

Does a heat pump use more electricity than baseboard heat?

Less, by a wide margin. Baseboard heaters turn one kWh into one kWh of heat, a COP of 1. A heat pump averaging COP 2.5 delivers the same heat for 40% of the electricity. In the coldest hours, a heat pump at COP 1.75 still uses 57% as much as baseboard. Only when backup resistance strips take over does use match baseboard, and only for those hours.

Will a heat pump double my electric bill?

It depends on what heated the house before. Replacing gas or oil, a heat pump adds new electricity use: in the Chicago example, about 6,900 kWh a year on top of everything else the house already uses. Replacing baseboard heat, total electricity use falls. The useful comparison is the heat pump's kWh cost against the fuel it replaces; our heat pump vs furnace operating cost guide shows the break-even COP method.

How do I find my heat pump's seasonal COP?

Divide the HSPF2 on the AHRI certificate or spec sheet by 3.412. Older equipment rated before 2023 shows HSPF instead of HSPF2; HSPF values are higher for the same equipment because the older test was less demanding, so do not mix the two. For cold-climate performance, look up the model's COP at 5F and 17F in the NEEP cold-climate heat pump list or the manufacturer's submittal.

Is a heat pump cheaper to run in Quebec or Ontario?

On these 2026 rates, Quebec's second-tier 11.142 cents is below Ontario's 13.2-cent upper tier, and Hydro-Quebec's energy price already includes delivery, while Ontario adds distributor delivery charges on top of the commodity price. An Ontario home on time-of-use or ultra-low overnight that runs the heat pump largely off-peak can narrow the gap.

How much electricity does a heat pump use per hour?

At full output, a 3-ton heat pump delivering 36,000 Btu/h at COP 2.5 draws about 4.2 kW, or 4.2 kWh in an hour. Most hours it runs at part load and uses less. Annual totals, not hourly peaks, decide the bill; hourly peaks decide whether your electrical service and panel can carry the load.