12,000 Btu/h per ton by definition. DOE on why square feet per ton oversizes
A ton is 12,000 Btu per hour by definition, so converting a load to tons is one division. The reason this page exists is what happens next: equipment comes in half-ton steps, so the answer is almost never a size you can buy, and rounding up is not the safe direction. Cooling is two jobs, and moisture only leaves the air while the coil is running. An oversized unit reaches the thermostat first and leaves the house cold and damp.
Cooling load to equipment size
Btu/h
sq ft
sq ft/ton
tons
This converts a load into tons. It does not produce the load: that takes a real Manual J, and the BTU calculator here does the conduction half of one.
Specification summary
Generated from on .
Reopen that address to reproduce these figures exactly.
Entered
Result
The load, and the sizes you can actually buy
Worked example
A sensible cooling load of 24,000 Btu/h at a sensible heat ratio of 0.75, on an 1,800 sq ft house, equipment in half-ton steps, SEER 16.
Total load. 24,000 ÷ 0.75 = 32,000 Btu/h. The extra 8,000 is moisture.
In tons. 32,000 ÷ 12,000 = 2.67 tons.
Sizes either side. 2.50 tons is 6.2% under. 3.00 tons is 12.5% over.
Heat moved at 3 tons. 36,000 Btu/h.
Electrical draw at SEER 16. 2,250 W, which is nothing like 36,000 Btu/h in watts.
Implied rule of thumb. 1,800 ÷ 2.67 = 675 sq ft per ton, derived rather than assumed.
That last figure is the interesting one. The rules of thumb in circulation run 400 to 600 square feet per ton, and this house works out at 675, which means every one of them would have oversized it. The rule is not a bad approximation of a good answer, it is a different calculation that happens to produce numbers in the same range. And note the electrical figure: 36,000 Btu/h of heat moved for 2,250 W drawn, because a heat pump moves heat rather than making it.
The formula
Two divisions, and the first one is the one people skip:
total = sensible / SHR tons = total / 12,000
sensible
the temperature part of the load, Btu/h. What a conduction calculation gives you
SHR
sensible heat ratio. The share of capacity that is temperature rather than moisture
12,000
Btu/h in a ton of refrigeration. A definition
step
equipment size increment, usually half a ton. The answer has to land on one
Dividing by SHR rather than multiplying is the part to be careful about. A sensible load of 24,000 at an SHR of 0.75 needs 32,000 Btu/h of total capacity, not 18,000. The equipment has to do both jobs and the rating describes the total.
Why rounding up is the wrong instinct
With almost everything on this site, ordering a bit extra is the safe direction. A spare gallon of paint, a spare sheet of board, a quarter yard more concrete: the cost of being short is always worse than the cost of a small surplus.
Cooling equipment is the exception, and it is worth understanding why rather than just being told.
A coil does two things at once. It lowers the air temperature, and once its surface drops below the dew point it condenses water out of the air. Both happen only while it is running. The thermostat, however, only measures the first one.
So an oversized system pulls the temperature down fast, satisfies the thermostat and stops, having barely begun the moisture work. The house reaches 72 degrees and stays damp, and the occupant turns the thermostat down further, which makes it worse. The failure mode of an oversized air conditioner is not waste, it is discomfort that more capacity cannot fix.
What a ton is, and what it is not
A ton of refrigeration is 12,000 Btu per hour, by definition, from the heat needed to melt a short ton of ice in a day. It has nothing to do with the weight of the equipment.
It is a rate of heat moved, and that is the distinction people trip on. 12,000 Btu/h is about 3,517 watts of heat shifted out of the house. The electricity the compressor draws to do it is far less, because a heat pump moves heat rather than making it, and the ratio between the two is the efficiency rating.
So a 3 ton system is not a 10.5 kW electrical load. It moves 36,000 Btu/h and draws whatever its efficiency requires to do so, which is a different number entirely and the one that matters for the electrical side.
And equipment comes in half-ton steps. Which means the answer to a load calculation is almost never a size you can buy, and rounding is a decision: up, and you are into the oversizing problem above; down, and you have a system that runs continuously on the hottest afternoon, which is what it was designed for.
Half a cooling load has nothing to do with temperature
Heating is one job. Cooling is two, and the second one is the reason oversizing an air conditioner fails rather than merely wasting money.
Sensible load is heat you can measure with a thermometer. Latent load is moisture, and removing it takes energy that never shows up as a temperature change at all: roughly 1,000 Btu per pound of water condensed out of the air.
Cooling and drying are separate jobs. A coil removes heat from the air and, once its surface is below the dew point, water as well. Both happen only while the coil is running.
An oversized system stops before it has dried anything. It pulls the temperature to setpoint quickly, satisfies the thermostat and shuts off. The sensible job is done and the latent job is barely started.
Which is why the failure is a cold, clammy house. Not a warm one. A house that feels damp at 72 degrees usually has too much equipment rather than too little, and no amount of extra capacity fixes it.
And why short cycling is the symptom to look for. Frequent short runs are what oversizing produces, and they are visible without any instrument.
Sensible heat ratio is the share of a system's capacity that goes to lowering temperature rather than removing moisture. It is a property of the equipment at a stated condition, not a constant, and it appears on the expanded performance data rather than the box.
Which is why this page asks for it rather than assuming, and why the sensible figure a conduction calculation produces is not the whole answer even when it is exactly right.
The size you calculate is not a size you can buy
Residential equipment comes in half-ton steps: 1.5, 2, 2.5, 3 and so on. A load calculation produces a number like 2.67 tons, which is not on the list.
So every job ends in a rounding decision, and the calculator prints both options with the percentage each is off by, because that is the information the decision actually needs.
The size down runs longer on the hottest afternoon. That is not a fault: design conditions are deliberately exceeded a few hours a year, and a system running continuously on the worst day is a system that is correctly sized. It also dehumidifies better, because it runs longer.
The size up idles. On a mild day it satisfies the thermostat in minutes and stops, which is where the moisture problem and the short cycling live.
Variable capacity equipment changes this calculation entirely, because it can modulate down to match a part load rather than cycling. That is a different product decision and not a sizing one.
A ton of cooling is not a ton of electricity
Twelve thousand Btu per hour is about 3,517 watts of heat moved. It is not 3,517 watts of electricity drawn, and the gap between those two is the entire point of a heat pump.
At SEER 16 a three ton system moving 36,000 Btu/h draws roughly 2,250 W. That is about 16 watts of heat shifted for every watt consumed, which is why cooling is cheap to run relative to what it achieves and why resistance heating is expensive relative to the same.
The practical consequence is on the electrical side. Sizing a circuit or a generator for an air conditioner means using the electrical draw from the nameplate, not converting the tonnage. Converting the tonnage overstates it by more than an order of magnitude.
The watt calculator on this site does that conversion and shows the definition behind it.
What this page does not do
It does not produce a load. It converts one. If you arrived with floor area and nothing else, the rule-of-thumb mode exists to show you what that method produces, and the page argues against using it.
It does not select equipment. That is ACCA Manual S, which matches specific equipment to a load at design conditions and accounts for how the capacity splits between sensible and latent at those conditions. It is a separate standard and it is not free.
It does not size ducts. Manual D, also not free. A correctly sized system on undersized ducts performs like an incorrectly sized one.
And it cannot see your ducts' location. Ductwork running through an unconditioned attic loses a substantial fraction of the capacity before it reaches a room, and that loss is part of a real load calculation rather than a correction applied afterwards.
Square feet per ton is the same mistake wearing a different unit
The cooling version of the rule of thumb is square feet per ton: pick a number between 400 and 600, divide, and buy that. It has the same defect as the heating version and the same source objects to it:
“Rules of thumb” are too often used to size comfort systems, which results in excessively oversized systems. This leads to increased cost, wasted energy, and too-frequent on and off cycling, which can lead to comfort and efficiency issues. A trained contractor or qualified designer will determine a home's specific heating and cooling loads to establish the right size equipment.
The reason it is worse for cooling is the latent half. A square-foot rule cannot see glazing, which is where solar gain enters and which is almost entirely a cooling load. It cannot see how many people are in the house, and people are a latent load. It cannot see whether the ducts run through a hot attic.
So the calculator on this page takes a load in Btu per hour, from a calculation rather than a rule, and converts it. If all you have is floor area, the honest position is that you do not yet have a load.
Pacific Northwest National Laboratory for the US Department of Energy
An official US government publication. It is quoted here because it is the clearest statement, from a body with no product to sell, that the square-foot method this page refuses to use is the problem rather than the shortcut.
Air Conditioning Contractors of America, described by BASC
The standard itself is not free and is NOT reproduced here. This page carries the conduction term only and names everything Manual J adds that it does not.
The design temperature difference is not published here either. Winter design temperature is a local value, it appears in IRC Table R301.2(1) as a jurisdiction-by-jurisdiction figure, and the calculator asks you for it rather than assuming a climate.
Frequently asked questions
How many tons of AC do I need?
Divide your total cooling load in Btu per hour by 12,000. The load has to come from a calculation rather than floor area: a sensible load of 24,000 Btu/h at a sensible heat ratio of 0.75 is a total of 32,000, which is 2.67 tons, and equipment comes in half-ton steps so the real choice is 2.5 or 3.
How many BTU is a ton of air conditioning?
12,000 Btu per hour, by definition, from the heat needed to melt a short ton of ice in 24 hours. So 2 tons is 24,000 Btu/h and 3 tons is 36,000. It describes heat moved, not electricity drawn.
How many square feet per ton of air conditioning?
There is no right answer, which is the problem with the question. The rules in circulation run 400 to 600 sq ft per ton, and the US Department of Energy's building science programme names that practice as the cause of excessively oversized systems. Work out the load and divide, and you get whatever your house actually is: on the worked example here, 675 sq ft per ton.
Is it better to oversize or undersize an air conditioner?
Neither, but oversizing fails worse. A coil only removes moisture while it runs, so an oversized system satisfies the thermostat and stops before it has dehumidified, leaving a cold damp house that more capacity cannot fix. An undersized system runs continuously on the hottest afternoon, which is what design conditions assume anyway.
What is sensible heat ratio?
The share of a system's capacity that goes to lowering temperature rather than removing moisture. A sensible load of 24,000 Btu/h at an SHR of 0.75 needs 32,000 Btu/h of total capacity, because the other 8,000 is doing the moisture work. It is a property of the equipment at stated conditions and appears in the expanded performance data rather than on the box.
How many watts does a 3 ton AC use?
Far less than the heat it moves. Three tons is 36,000 Btu/h of heat shifted, and at SEER 16 the draw is around 2,250 W. Do not size a circuit by converting tonnage to watts: use the electrical figures from the nameplate.
Should I round up to the next size?
Usually not. The next half-ton step is often more than 10% over the load, and on cooling that is the direction that causes problems. The size down runs longer on the hottest day, which is what it is designed for, and dehumidifies better because it runs longer.
Check these numbers yourself
Proper Sizing HVAC System, Building America Solution CenterPacific Northwest National Laboratory for the US Department of Energy. An official US government publication. It is quoted here because it is the clearest statement, from a body with no product to sell, that the square-foot method this page refuses to use is the problem rather than the shortcut.
ACCA Manual J, Residential Load CalculationAir Conditioning Contractors of America, described by BASC. The standard itself is not free and is NOT reproduced here. This page carries the conduction term only and names everything Manual J adds that it does not.
How every figure here is verified: Sources & Method. Who builds this: About. Found something wrong? Tell us and it gets fixed or removed.
Figures on this page last checked against the source documents on 2026-09-08. Codes are amended locally; confirm against the edition your jurisdiction enforces.
This converts a load into equipment sizes. It does not produce the load. That takes a real ACCA Manual J, which is not a free document and is not reproduced here. Equipment selection is Manual S and duct design is Manual D, also not free and also not here. Sensible heat ratio is a property of specific equipment at stated conditions and is an input rather than an assumption. Sources read 8 September 2026.