Heat Pump Noise Level: Decibel Ratings, Distance and Noise Bylaws

How Loud a Heat Pump Is, in Numbers You Can Check

A cold-climate heat pump's outdoor unit is rated at about 51 to 58 dB(A) of sound pressure in Mitsubishi's 2025 Hyper-Heat multi-zone submittals, 2 dB louder in heating than in cooling. Toronto's noise bylaw caps a residential air conditioner or similar device at 45 dB(A) overnight and 50 dB(A) by day, measured near the neighbour's building, so the spec number alone never settles it. What matters is how much of that number survives the trip across the side yard.

That trip is arithmetic. For a small outdoor unit in open air, sound pressure drops about 6 dB each time the distance doubles. A unit rated 58 dB(A) at a 1 metre reference is near 46 dB(A) at 4 metres before any reflections are counted, near 49 dB(A) once the house wall behind it reflects sound toward the neighbour, and near 52 dB(A) in an inside corner. The difference between passing and failing a 45 dB(A) night limit is often one wall and a few metres.

The Two Decibel Figures on a Spec Sheet

Outdoor-unit literature uses two different quantities, and mixing them up is the most common reason homeowners and contractors talk past each other.

Sound pressure level (Lp) is what a microphone reads at one spot. It depends on distance, on the room or yard the unit sits in, and on where the microphone is. The Mitsubishi figures above are sound pressure levels. The submittal gives the operating conditions (cooling at 95°F outdoors, heating at 47°F outdoors) but does not print the measurement distance on the page, which means the number tells you how two Mitsubishi units compare with each other and not, by itself, what a neighbour hears.

Sound power level (Lw) describes the total acoustic energy the unit radiates, independent of distance. Some data sheets give sound power instead of, or as well as, sound pressure, and a sound power figure always looks higher than a sound pressure figure taken a metre or more from the same machine, for reasons that are geometry, not loudness. A sound power value has to be converted before it can be compared with a bylaw limit. The conversion, for a unit sitting on the ground in open air, is:

Lp = Lw + 10 × log10(Q ÷ (4 × π × r²))

where r is the distance in metres and Q is the directivity factor: 2 for a unit on open ground (half the sphere is blocked by the ground), 4 for a unit on the ground against one wall, 8 for one tucked into an inside corner. At 1 metre with Q = 2, that formula subtracts 8 dB from the sound power figure. Against a wall (Q = 4) it subtracts 5 dB. So a condenser with a sound power of 72 dB(A) produces about 64 dB(A) one metre away in the open and about 67 dB(A) one metre away against the house.

A-weighting is the "(A)" in dB(A). It discounts low frequencies roughly the way human hearing does at moderate levels. Every bylaw in the table further down is written in A-weighted decibels, which is convenient for comparison and also a blind spot: a compressor hum or a fan's blade-pass tone can be annoying through a closed window while adding very little to the A-weighted total.

A practical consequence: when a contractor says "this unit is 56 decibels", ask which of the three they mean. Sound pressure at what distance and in which mode, or sound power. The answer changes the setback by a factor of two or three.

Published Ratings for Cold-Climate Outdoor Units

The table lists the outdoor-unit sound figures from three current manufacturer submittals read on November 13, 2026. They are rated-condition numbers from one brand. Each manufacturer tests its own way, so a figure from another brand's sheet is less comparable with these than these are with each other.

All three rows come from Mitsubishi Electric's US submittals coded METUS_2025-03. Cooling is rated at 95°F outdoors and heating at 47°F outdoors; neither sheet prints a measurement distance or a sound power level. A 2022 Mitsubishi Electric Sales Canada submittal for an older three-zone model, the MXZ-3C30NAHZ3-U1, lists 54 dB(A) in cooling and adds a line the US sheets leave out: "Rated data is determined at a fixed compressor speed." That sentence is the key to reading every row above. The rating is one speed, chosen by the test procedure, not the loudest speed the inverter can reach.

Three patterns are worth reading out of the table directly:

• Heating is louder than cooling in every row. In each of these submittals the heating figure is 2 dB above the cooling figure. In a cold climate the heating figure is the one that matters, because the unit runs longest, at the highest compressor speed, on winter nights when bedroom windows face it and background traffic noise is lowest.
• The same box can be 5 dB louder. All three units share a 950 x 330 x 796 mm cabinet, yet the three-zone models are rated 5 dB above the two-zone model in both modes. Between the two three-zone models the rating does not move at all, even though the larger one is rated for 3,600 Btu/h more heating. By the distance arithmetic below, 5 dB is the difference between a unit 4 metres from a window and one about 7 metres away. Cabinet size says nothing about sound; the submittal line does.
• The rated point is not the loudest point. Heating ratings are taken at 47°F (8.3°C). On a -15°C night the same inverter compressor runs much faster to hold capacity, and the fan follows it. No submittal in the table publishes a sound level at low ambient temperature, so the rated figure is a floor for winter operation, not a ceiling.

If a quote names a unit that is not in the table, the same numbers are usually on page one or two of its submittal, in the outdoor-unit column labelled "Sound Pressure Level" or "Sound Level". Ask for that page with the quote. Our cold-climate heat pump performance data page covers the capacity side of the same submittals.

How Sound Falls Between the Unit and the Lot Line

For distances that are large compared with the unit itself, an outdoor unit behaves like a point source, and sound pressure falls with the square of distance. In decibels that becomes:

Lp(r) = Lp(r0) − 20 × log10(r ÷ r0)

With a 1 metre reference (r0 = 1), the drop at each distance is:

Every value in that table is 20 × log10 of the distance, worked by hand and rounded to one decimal. Halving the distance adds the same 6 dB back.

Four assumptions sit under that table, and each one can move the answer:

1. The reference distance. The table starts from a level at 1 metre. If a manufacturer's figure was measured at some other distance, the start point shifts. Where a data sheet does not say, treat the 1 metre reading as an assumption and run the numbers a second time with the sound power method if a sound power figure is available.
2. Near-field error. A 950 mm wide outdoor unit is not a point at 1 metre; its fan and compressor sit in different places on the cabinet. Close in, the measured level falls more slowly than 6 dB per doubling. The rule is reliable once you are two or three cabinet widths away, which on a city lot is usually the neighbour's wall anyway.
3. Reflections add, they do not subtract. A hard wall directly behind the unit sends sound that would have gone into the house back out toward the yard. In the directivity terms above, moving from open ground (Q = 2) to against one wall (Q = 4) raises every distance's level by 3 dB, and an inside corner (Q = 8) adds another 3 dB. A narrow side yard between two masonry walls can act like a channel and lose less than the table shows.
4. Nothing in the way. The table assumes open air. A solid fence, a garage or the house corner in the line of sight cuts the level at the window behind it. That cut depends on height, distance and frequency, and we do not put a single number on it here.

Setback distances that meet each limit

Turning the formula around gives the distance at which a unit falls to a given limit: r = 10^((L + reflection − limit) ÷ 20), where L is the rated level at 1 metre. The table below uses that formula for a range of rated levels, for a unit on open ground with one reflecting wall behind it (+3 dB) and in an inside corner (+6 dB):

For ducted condensers rated in sound power, the same exercise runs through the Lp formula from the previous section, r = √((Q ÷ 4π) × 10^((Lw − limit) ÷ 10)):

Read the two tables together and the difference between equipment types is plain. A unit rated in the mid-40s dB(A) at 1 metre (the top rows of the setback table) meets a 45 dB(A) night limit within a couple of metres even with a wall behind it. A large ducted condenser in the 72 to 76 dB(A) sound power range needs 9 to 20 metres to do the same, which is more space than most semi-detached and row-house side yards have. Those are steady-state, rated-condition numbers; the next section covers the moments that run hotter.

Adding Units, and Subtracting the Street

Decibels do not add like ordinary numbers, because they are logarithms. To combine two sources:

L_total = 10 × log10(10^(L1 ÷ 10) + 10^(L2 ÷ 10))

Two equal sources add 3 dB, not double: two units each producing 45 dB(A) at a neighbour's window give 48.0 dB(A) together. A source 6 dB quieter than another adds about 1 dB: 48 dB(A) plus 42 dB(A) totals 49.0 dB(A). This matters on a lot with a heat pump and an existing air conditioner, a heat pump water heater's outdoor unit, or two neighbours' units facing the same window.

The same formula run backwards separates a heat pump from background noise when you measure it. If the meter reads 48 dB(A) with the unit running and 44 dB(A) with it off, the unit's own contribution is:

10 × log10(10^4.8 − 10^4.4) = 45.8 dB(A)

That correction only works when the unit is at least about 3 dB louder than the background. When the two readings are within 1 or 2 dB, the background is doing most of the work and the subtraction becomes unreliable.

Why the bylaw's "one hour" matters

Toronto's and Ontario's limits are written as an equivalent continuous level over one hour (Leq, 1 h), not as a peak. Leq is an energy average. A unit that runs at 50 dB(A) at the window for 30 minutes of an hour, with 35 dB(A) of background the rest of the time, has an hourly level of:

10 × log10(0.5 × 10^5.0 + 0.5 × 10^3.5) = 47.1 dB(A)

Cycling helps less than people expect, because the loud half dominates the average. An inverter unit that runs at low speed all hour may produce a lower Leq than a single-stage unit that cycles on and off at full output, even though the inverter is never silent.

The Limits That Apply, by Jurisdiction

The table covers three documents, two from Ontario and one from Oregon. Each entry gives the place the level is measured, which differs between codes and changes the answer as much as the dB value does.

Toronto: a dedicated air conditioner clause

Toronto is unusual in naming residential air conditioners at all. Section 591-2.8 D sets the 45 and 50 dB(A) limits for "a residential air conditioner or similar residential device". The clause does not name heat pumps, but a heat pump's outdoor unit is the same machine run in both directions, so "similar residential device" is the clause a bylaw officer would most plausibly reach for. Two details in the same section shape real disputes:

• Ambient override. Where the hourly ambient level already exceeds the 45 or 50 dB(A) limit, the limit becomes the ambient level itself. On a lot next to a busy arterial road, the unit is allowed to be as loud as the traffic, but no louder.
• Where the microphone goes. The point of reception is on the receiving property, near the facade of a building, 1.5 metres above ground. It is not at the lot line. On a 1.2 metre side yard between two houses, the neighbour's facade may be 2 or 3 metres from the outdoor unit; on a corner lot it may be 15.

The same chapter's definition of "stationary source" expressly excludes residential air conditioners, so a homeowner's heat pump is judged under the device clause, not the general stationary-source clause that applies to businesses.

Ontario: the provincial guideline leaves homes to municipalities

NPC-300 is the guideline Ontario uses for approving industrial and commercial noise sources and for land-use planning near them. Its Class 1 limits for urban areas are 50 dB(A) outdoors and at the plane of a window during the day and evening, and 45 dB(A) at the plane of a window overnight. But NPC-300 lists "residential air conditioning devices including air conditioners and heat pumps" among sources that are not stationary sources. The ministry's older residential air conditioner publications, NPC-216 (1993) and a 1994 installation guideline, are cited in NPC-300's reference list; enforcement against a neighbour's unit runs through the municipal bylaw, which is why Toronto wrote its own clause. Other Ontario municipalities write their own bylaws, and they differ, so check the local text before relying on the Toronto numbers outside Toronto. Our Ontario heat pump cost guide covers the price and program side for the same homes.

Portland: one table for every source, read at the lot line

Portland, Oregon does not single out HVAC equipment. Its code sets a single grid of maximum levels by the zone the sound comes from and the zone it lands in, measured at the property line. Residential to residential is 55 dBA during day hours (7 a.m. to 10 p.m.) and 5 dBA lower, 50 dBA, at night. Measuring at the property line instead of at the neighbour's facade makes Portland's 50 dBA night limit harder to meet than Toronto's 45 dB(A) in a narrow side yard, even though the number is higher: a unit 1.5 metres from the lot line gets only 1.5 metres of distance loss, while the Toronto microphone sits at the neighbour's wall.

A Side-Yard Case, Worked Twice

Take a detached house in Toronto with a 1.2 metre side yard, and a neighbour whose bedroom window faces it. The installer proposes putting the MXZ-3D24NLHZ against the house wall, centred in the side yard. The neighbour's facade, where Toronto's point of reception sits, is 2.5 metres from the front of the cabinet.

Inputs, with the assumptions labelled:

• Rated heating level: 58 dB(A), from the submittal, treated as a 1 metre free-field value (assumption: the submittal does not state the distance).
• Reflections: the house wall behind the unit and the ground, +3 dB over the rating's condition (assumption: Q goes from 2 to 4); the neighbour's wall across the yard adds reflected sound too, which this estimate ignores, so the real figure is likely higher.
• Distance loss to 2.5 m: 20 × log10(2.5) = 8.0 dB.
• Running pattern: continuous operation through a cold hour, so Leq equals the running level.

Estimated level at the neighbour's facade: 58 + 3 − 8.0 = 53 dB(A). That is 8 dB over Toronto's 45 dB(A) night limit and 3 dB over the 50 dB(A) day limit, at the rated heating point, before the colder-night speed increase.

Now move the same unit to the back of the house, on a stand against the rear wall, 9 metres from the nearest neighbour facade, with nothing but open yard between:

• Distance loss to 9 m: 20 × log10(9) = 19.1 dB.
• Estimated level: 58 + 3 − 19.1 = 41.9, about 42 dB(A).

That is 3 dB under the night limit at the rated point. On the coldest nights, with the compressor working harder than at 47°F, part of that margin will be used up, and the owner should expect the unit to be audible at the rear facade even if it is legal.

A third option, a single-zone unit with a lower rating serving only the main floor, changes the input rather than the distance. Whether that is possible depends on the house's heat loss, which is a sizing question covered in our heat pump guides. The location choice itself, including snow, drainage and service access at the rear, belongs to the outdoor unit placement guide; the acoustic test above is the extra check to run on each candidate spot.

When the Spec-Sheet Arithmetic Understates the Noise

The rated figure and the 6 dB rule describe one steady state. Several ordinary situations run louder or sound worse than the arithmetic suggests:

Low outdoor temperature. An inverter heat pump raises compressor speed as the outdoor temperature falls, to hold capacity. The submittal's 47°F heating rating describes a mild day. None of the three submittals gives a sound level at -15°C, which leaves the cold-night level as an unknown that sits above the rating. Leave a margin of several decibels when the calculated level is close to a limit.

Defrost. In heating, frost builds on the outdoor coil and the unit periodically reverses to melt it. The switch produces a hiss or whoosh from the reversing valve and refrigerant flow, the outdoor fan often stops, and the coil may steam. These events are short, so they move a one-hour Leq less than they move a neighbour's sleep. Our defrost cycle guide explains what is normal and what is a fault.

Tones and low frequencies. A-weighting discounts the low-frequency hum of a compressor. A unit can meet an A-weighted limit and still produce an audible drone through a closed window, especially through lightweight wall or window assemblies. Toronto's residential device clause adds no penalty for tonal sound, so this is a neighbour-relations problem more often than a legal one.

Structure-borne vibration. A wall bracket bolted to a bedroom wall can put vibration into the framing, and the room on the other side then hears a hum that never shows up in an outdoor measurement. Isolation mounts and keeping line sets from touching framing are covered in the placement guide.

Ageing and faults. A bent fan blade, a loose panel, a failing fan motor bearing or ice striking the fan all add noise that no rating covers. These are repair items, not design items.

Night background. A street that reads 55 dB(A) at 5 p.m. may read 35 dB(A) at 3 a.m. A unit that is inaudible during a daytime site visit can be the loudest thing in the neighbourhood at night. Toronto's ambient override works both ways: it allows more noise beside a busy road, and it gives no room at all on a quiet one.

Measuring a Unit Yourself

A homeowner's measurement will not stand in for a bylaw officer's calibrated meter, but it tells you whether a concern is plausible before anyone files a complaint or spends money on a fence.

1. Use an A-weighted meter or app set to dB(A) and, if it has one, an Leq or average mode. A spot reading of the maximum overstates the hourly level.
2. Stand where the code measures. In Toronto that is near the receiving building's facade, about 1.5 metres up. In Portland it is the property line. Measuring at the unit tells you nothing useful about compliance.
3. Measure on a cold night with the unit in heating, ideally during a long run, not in the first few minutes after start-up and not during a defrost.
4. Measure the background too. Ask the owner to switch the unit off for a few minutes, or wait for a cycle, and record the level without it. Subtract on a log basis as shown above.
5. Write down the outdoor temperature, the time, and the distance. A level without its conditions cannot be compared with a limit or with a later reading after a fix.

Phone microphones vary, and their low-frequency response is weak. Treat a phone reading within 3 dB of a limit as "too close to call", not as a pass or a fail.

Lower Levels Before and After Installation

Before the contract is signed, the cheapest decibels come from equipment choice and location:

• Ask for the outdoor unit's submittal sheet with the quote, and compare heating sound levels between the bids. A two-unit design, with a smaller outdoor unit for each part of the house, can sometimes replace one large multi-zone unit at a lower level per unit. Our heat pump quote comparison worksheet has room to record it.
• Put distance between the unit and the nearest bedroom window you do not own. Each doubling is 6 dB. On many lots the rear wall buys more than any product upgrade does.
• Avoid inside corners and narrow channels between hard walls. Each added reflecting surface takes back about 3 dB.
• Do not oversize the outdoor unit. A larger outdoor unit than the house needs carries a higher rating, and an inverter unit that spends most of the winter well below its maximum speed is quieter for most hours.

After installation, the options are more expensive and less certain:

• A barrier that blocks the line of sight between the unit and the window. It has to be solid, without gaps, and taller than the line between the top of the unit and the window to do much; a lattice or hedge does very little. It must also respect the manufacturer's airflow clearances, or it will cost efficiency and capacity.
• A quieter-mode setting, if the model's controls offer one. A setting that caps outdoor-unit speed at night lowers the level and the capacity at the same time, and on a cold night that trade may not be acceptable. Check the owner's manual for the specific model rather than assuming the feature exists.
• Moving the unit. Relocating an outdoor unit means new line sets, electrical work and a pad or stand. It costs a fraction of a new system and sometimes solves what nothing else can.

An enclosure or box around the unit is the option to treat most cautiously: an enclosure that is not designed for the specific model can choke airflow, recirculate cold discharge air in winter, and risk snow.

If a Neighbour Complains, or You Are the Neighbour

Most disputes over heat pump noise end without a bylaw officer when both sides look at the same numbers. A useful sequence is:

1. Find the unit's model number on its rating plate and look up its submittal sound levels.
2. Measure the distance from the unit to the receiving facade or lot line, depending on the local code, and run the arithmetic in this guide.
3. Take a measurement on a cold night following the five steps above.
4. If the estimate and the measurement both come in under the local limit, the problem is audibility, not legality, and the fixes are goodwill items: a barrier, a night setting, a mount check.
5. If either comes in over the limit, the installer is the first call, because location and mounting are part of the installation. A municipal complaint is the last step, not the first.

In Toronto, the residential device clause, not the general stationary-source clause, is the one that applies, so a complaint that quotes the wrong section starts on the wrong foot. In Portland, the question is the level at the property line, which an installer can check before installation from the setback alone.

For readers still choosing a system, our heat pumps hub collects the cost, sizing and rebate guides in one place. Bring the outdoor unit's sound figure and the distance to the nearest neighbour's window to the first site visit, because those two numbers settle most noise disputes before any equipment is ordered.

Frequently Asked Questions

How many decibels is a heat pump outdoor unit?

The Mitsubishi Hyper-Heat multi-zone units in the ratings table above are rated 51 to 56 dB(A) in cooling and 53 to 58 dB(A) in heating, at the manufacturer's test conditions. Indoor heads are a separate, much quieter figure and do not count toward a neighbour's limit.

Is 58 dB(A) too loud for a heat pump next to a neighbour?

Not by itself. At 58 dB(A) from a 1 metre reference with one reflecting wall, the level falls to 45 dB(A) at about 6.3 metres and to 50 dB(A) at about 3.5 metres. A unit that close to someone else's bedroom window can break a 45 dB(A) night limit; the same unit at the far side of a yard usually will not.

Does a heat pump get louder in winter?

At the rated points, yes: each Mitsubishi submittal in the ratings table rates heating 2 dB above cooling. On cold nights an inverter compressor runs faster than at the 47°F rating point, and defrost cycles add short bursts of noise, while traffic and other background sound is often at its lowest.

How far from the property line does a heat pump have to be?

None of the three codes summarised here sets a fixed setback distance for heat pumps. They set a sound level instead: Toronto at the neighbour's facade, Portland at the property line. The distance follows from the unit's rating, using the setback tables in this guide, and your municipality's zoning rules may add their own siting limits.

Will a fence stop heat pump noise?

A solid fence with no gaps that blocks the line of sight between the unit and the window lowers the level behind it. Lattice, slatted fences and hedges do little. Any barrier must keep the manufacturer's airflow clearances, or the unit loses capacity and efficiency and may get louder as it works harder.

Who enforces heat pump noise limits?

The municipality that wrote the bylaw. In Toronto that is the city's bylaw enforcement under Chapter 591. In Portland it is the city under Title 18. Ontario's NPC-300 guideline does not cover a homeowner's heat pump, because it lists residential air conditioning devices, heat pumps included, as not being stationary sources.