A reference of 26 statistics on how often home heating systems fail badly enough to leave a household uncomfortably cold for 24 hours or more: boilers, furnaces, heat pumps and space heaters compared, the same comparison within each region, building type, home age and owners versus renters, apartment buildings, heating fuel, and how often the heat breaks more than once. Every figure is our own computation from the U.S. Census Bureau's 2023 American Housing Survey public use file (census.gov), checked against the Bureau's published totals, with margins of error from the survey's replicate weights.
Key statistics
6 highlights from this report
Key statistics
Key takeaways
In the winter of 2022-23, boilers left households cold more often than any other central heating system, and the gap is widest where the household does not control the boiler: in large apartment buildings and among renters. Households with ducted heat pumps were no more likely than furnace households to be left cold by a breakdown.
Boiler homes went cold from a heating breakdown 2.3 times as often as furnace homes: 5.3% vs 2.3%.
About 1 in 10 households (10.4%) in boiler-heated buildings of 5+ apartments were left cold for a day or more by a breakdown.
Renters with boilers were left cold by a breakdown 3.4 times as often as owners with boilers: 9.4% vs 2.7%.
Heat pump households were no more likely than furnace households to be left cold by a breakdown: 1.9% vs 2.3%.
Matched on region, building, home age and tenure, boiler households were still left cold about 1.6 times as often (5.3% vs 3.4%).
57% of boiler households left cold by a breakdown had two or more breakdowns of 6+ hours that winter, vs 41% with furnaces.
How we built this report
Every survey figure on this page is a computation we ran in October 2026 against the raw federal survey file, not a number copied from another publication.
- Downloaded the public use microdata
We used the national public use file of the 2023 American Housing Survey, which the U.S. Census Bureau runs for the Department of Housing and Urban Development. Households were interviewed between May and October 2023; each names its main heating equipment and fuel, and says whether the home was uncomfortably cold for 24 hours or more the previous winter (2022-23) and why.
- Defined one measure
A heating breakdown here means the household was uncomfortably cold for 24 hours or more in winter 2022-23 because the main heating equipment broke down. The base is the 46,548 sampled households that had heating equipment and lived in the home that winter, weighted to 127.6 million.
- Validated against Census figures
Our weighted counts match the Census Bureau's own verification figures to the thousand: of all occupied homes, 90.0 million are heated mainly by a furnace, 11.0 million by a steam or hot-water system, 15.1 million by a heat pump, and 3.41 million households left cold by a heating breakdown.
- Margins from replicate weights
Each percentage carries a 90% margin of error from the survey's 160 replicate weights, the method the Census Bureau recommends. Where a difference is within its margin of error, we say so.
- Matched comparison
To estimate how much of the gap comes from the building rather than the system, we reweighted furnace households to match boiler households (and, separately, heat pump households) on region, building type, home age in three bands and owners versus renters, then compared rates. The match is coarse: it narrows the comparison but cannot isolate the equipment.
- Second review
Written by one co-founder, reviewed by the other before publishing.
These are breakdowns households reported for their own home, not repair records, and they count only failures that left the home uncomfortably cold for a day or more, not every service call. The survey records the type of the main heating system but not its age, brand or maintenance history, so the comparisons describe the systems as they exist in US homes today, not a lab test of the equipment. The winter in question is 2022-23, the one before each household's 2023 interview. Heat pumps here are centrally installed, ducted systems, as the Census Bureau defines them; ductless mini-splits fall under other categories, and many ducted heat pumps have backup electric heat that can keep a fault from leaving a home cold. Renters include households living without paying rent, as in Census tables. Rates for small groups, such as heat pump homes in the Northeast and Midwest, rest on few cases and carry wide margins. The 2023 survey's response rate was 59.2%, and margins of error cover sampling error only.
Heating breakdowns in US homes, by the numbers
All 26 figures, grouped by question, each computed from the 2023 American Housing Survey and shown with its 90% margin of error where it matters.
Breakdowns by type of heating system
According to the U.S. Census Bureau's 2023 American Housing Survey (census.gov), 3.4 million US households were uncomfortably cold for 24 hours or more in winter 2022-23 because their main heating equipment broke down. Split by the type of system, the rate goes from 2.3% for furnaces to 5.3% for boilers.
Households left cold 24+ hours by a heating breakdown, winter 2022-23
By main heating system. The thin line on each bar is the 90% margin of error.
Households with heating equipment that lived in the home in winter 2022-23 (127.6 million), by main heating equipment. Source: RapidEye Research analysis of the 2023 American Housing Survey.
Statistic 1
3.4 million households, 2.7% of those that lived in their home in winter 2022-23, were uncomfortably cold for 24 hours or more because the main heating equipment broke down.
RapidEye Research analysis of the 2023 American Housing Survey (matches the Census Bureau's published total)
Statistic 2
Boilers (steam or hot-water systems): 5.3% of households, about 560,000 of 10.6 million, were left cold for a day or more by a breakdown.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 3
Forced-air furnaces: 2.3% of households, about 2.0 million of 86.5 million.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 4
Electric heat pumps (ducted): 1.9% of households, about 270,000 of 14.5 million.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 5
Boiler homes went cold from a breakdown 2.3 times as often as furnace homes, a gap of 3.0 percentage points (±0.8).
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 6
Floor, wall and other pipeless furnaces: 3.9%; built-in electric baseboard or coils: 3.1%.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 7
2.3 million households rely on portable electric heaters as their main heat; 6.5% of them were left cold by a breakdown, and 48% of them rent.
RapidEye Research analysis of the 2023 American Housing Survey
What this means: two in three US households heat with a forced-air furnace, and that is the benchmark: about 1 in 44 was left cold for a day or more by a breakdown. Boilers sit well above it and heat pumps slightly below. For what the repair itself costs, see our HVAC cost statistics.
Is it the boiler or the building?
Boilers are not spread evenly. They are concentrated in older Northeast homes and in apartment buildings, where heat problems are more common for furnaces too. So we compared boilers and furnaces inside each region, building type, home age and tenure, and then matched the two groups on all four at once.
| Group | Furnace | Boiler | Heat pump |
|---|---|---|---|
| Region | |||
| Northeast | 2.8% ±0.5 | 5.3% ±1.0 | 1.5% ±1.1 |
| Midwest | 2.2% ±0.3 | 4.7% ±1.9 | 1.0% ±1.0 |
| South | 2.3% ±0.3 | 5.7% ±3.5 | 2.0% ±0.5 |
| West | 2.0% ±0.3 | 7.0% ±3.0 | 1.2% ±0.8 |
| Building | |||
| Houses | 2.0% ±0.2 | 3.0% ±0.8 | 1.9% ±0.4 |
| 2 to 4 apartments | 3.2% ±0.9 | 5.2% ±2.2 | 3.4% ±2.3 |
| 5 or more apartments | 3.2% ±0.5 | 10.4% ±2.0 | 1.3% ±0.9 |
| Year built | |||
| Before 1960 | 3.4% ±0.4 | 5.9% ±1.2 | 2.4% ±1.4 |
| 1960 to 1989 | 2.4% ±0.3 | 4.8% ±1.3 | 2.2% ±0.7 |
| 1990 or later | 1.5% ±0.2 | 3.8% ±2.2 | 1.5% ±0.5 |
| Tenure | |||
| Owners | 1.8% ±0.2 | 2.7% ±0.7 | 1.7% ±0.4 |
| Renters | 3.4% ±0.4 | 9.4% ±1.6 | 2.2% ±0.8 |
Households with heating equipment that lived in the home in winter 2022-23. Houses include attached single-family homes. Boiler rates in the South and West rest on about 220 sampled households each; the boiler-vs-furnace gaps in the South and in 2 to 4 apartment buildings are within their margins of error, and the boiler figure for homes built since 1990 rests on few cases. Source: RapidEye Research analysis of the 2023 American Housing Survey.
Statistic 8
75% of boiler households are in the Northeast, and 58% live in homes built before 1960.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 9
28% of boiler households live in buildings with five or more apartments, twice the furnace share (14%); 39% rent, against 30%.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 10
Matched on region, building type, home age and tenure, furnace households come to 3.4%, against 5.3% for boilers: about 1.6 times, a gap of 1.9 points (±1.0).
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 11
In houses alone, the boiler gap narrows to 3.0% against 2.0% for furnaces.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 12
Boiler households were left cold by a breakdown more often in homes of every age: 5.9% vs 3.4% in homes built before 1960, and 3.8% vs 1.5% in homes built since 1990, the newest group resting on few boiler cases.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 13
In the Northeast, where most boilers are, 5.3% of boiler households were left cold for a day or more by a breakdown, against 2.8% with furnaces.
RapidEye Research analysis of the 2023 American Housing Survey
What this means: matching accounts for nearly 40% of the raw boiler gap; the other 1.9 points remain. The raw gap is small in houses (3.0% vs 2.0%) and large in buildings with five or more apartments (10.4% vs 3.2%), which suggests shared central boilers, and how fast they are repaired, matter, not just boiler technology.
Apartments and renters
In an apartment building, a central boiler often heats every unit, and the tenants cannot repair it themselves. That is where the survey shows the largest gap.
Buildings with 5+ apartments
Houses
Renters
Owners
Statistic 14
In buildings with five or more apartments heated by a boiler, 10.4% of households, about 300,000, were left cold for a day or more by a breakdown: about 1 in 10.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 15
In buildings of the same size heated by furnaces, the rate was 3.2%: boiler buildings were 3.3 times as likely to leave a household cold.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 16
54% of the boiler households left cold by a breakdown live in buildings with five or more apartments.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 17
Renters with boilers were left cold by a breakdown 3.4 times as often as owners with boilers: 9.4% vs 2.7%.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 18
Among renters, boiler households were left cold by a breakdown 2.8 times as often as furnace households: 9.4% vs 3.4%.
RapidEye Research analysis of the 2023 American Housing Survey
What this means: for a landlord or building manager, a central boiler is a single point of failure for every unit it heats. State inspection rules for these boilers vary, including which small buildings are exempt. Our boiler inspection reference compares the rules in 21 jurisdictions.
Heat pumps vs furnaces
Heat pumps are often doubted as winter heat. In the 2023 survey, households that heat with a ducted heat pump were no more likely than furnace households to be left cold by a breakdown. The survey counts only centrally installed, ducted heat pumps; ductless mini-splits fall under other categories.
Statistic 19
1.9% of heat pump households were left cold for a day or more by a breakdown, against 2.3% with furnaces.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 20
Matched to heat pump households on region, building type, home age and tenure, furnace households come to 2.1%, against 1.9% for heat pumps: a difference within the margin of error.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 21
In no region were heat pump households left cold more often than furnace households: South 2.0% vs 2.3%, West 1.2% vs 2.0%, Northeast 1.5% vs 2.8%, Midwest 1.0% vs 2.2%. The regional gaps are within or near their margins of error.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 22
Among homeowners, 1.7% of heat pump households and 1.8% of furnace households were left cold for a day or more by a breakdown.
RapidEye Research analysis of the 2023 American Housing Survey
What this means: the survey gives no sign that ducted heat pumps leave households cold more often than furnaces. It measures cold homes, not equipment faults, and backup heat may cover some heat pump failures. Nearly eight in ten heat pump homes are in the South, and only seven sampled heat pump households in the Northeast and Midwest reported a breakdown, so those regional figures show the direction, not a precise rate.
By heating fuel
Fuel matters less than equipment. Gas and electric homes sit close to the national rate; fuel oil, which mostly feeds boilers (58% of oil-heated homes), sits highest.
Statistic 23
Fuel oil 3.9% (±1.1), piped gas 2.8%, electricity 2.5%, propane 2.4% of households left cold by a breakdown.
RapidEye Research analysis of the 2023 American Housing Survey
What this means: the fuel-oil figure carries a wide margin; read it as "highest, by a modest amount", not as a precise ranking.
Repeat breakdowns, and other reasons homes got cold
For many households the heat broke down more than once. And a breakdown is only one reason a home goes cold: the survey also asks about utility outages, undersized heating, poor insulation and the cost of fuel.
Statistic 24
Of the 3.4 million households left cold for a day or more by a heating breakdown, 46% had two or more breakdowns lasting six hours or longer that winter, and 22% had four or more.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 25
Repeat breakdowns were more common in boiler homes: 57% of boiler households left cold by a breakdown had two or more breakdowns lasting six hours or longer, against 41% of furnace households.
RapidEye Research analysis of the 2023 American Housing Survey
Statistic 26
10.0 million households (7.8%) were uncomfortably cold for 24 hours or more for any reason. Equipment breakdown was the most cited reason (34%), then utility interruptions (28%), heating too weak for the cold (16%), poor insulation (15%) and the cost of heating (9%).
RapidEye Research analysis of the 2023 American Housing Survey; households could give more than one reason
What this means: a broken system is the single biggest reason US homes go cold, and in boiler homes it is a bigger share still: 48% of cold boiler households blamed a breakdown, against 33% of cold furnace households. After-hours heating repairs carry their own premium; see our emergency repair cost statistics.
Cite this study
Academic or press use: copy a ready-made reference. RapidEye is the publisher.
Quick FAQ
Which heating system breaks down the most?
Among central systems, steam and hot-water systems (boilers) left households cold most often. In RapidEye Research's analysis of the 2023 American Housing Survey, 5.3% of households heated mainly by a boiler were uncomfortably cold for 24 hours or more in the winter of 2022-23 because the system broke down, against 2.3% with a forced-air furnace and 1.9% with a heat pump. Households relying on portable electric heaters as their main heat reported 6.5%.
Are boilers less reliable than furnaces?
In household reports, boilers left people cold more often, though part of the gap comes from where boilers are: three in four boiler households are in the Northeast and more than half live in homes built before 1960. Giving furnace households the same mix of region, building type, home age and owners versus renters raises the furnace rate to 3.4%, against 5.3% for boilers, about 1.6 times. In houses the gap is 3.0% vs 2.0%; in buildings with five or more apartments it is 10.4% vs 3.2%.
Do heat pumps break down more often than furnaces?
Not in this survey. In the 2023 American Housing Survey, 1.9% of households heated by a ducted heat pump were left cold for 24 hours or more by a breakdown in the winter of 2022-23, against 2.3% with a furnace. Comparing furnace households with the same mix of region, building type, home age and tenure as heat pump households gives 2.1%, a difference within the margin of error. In no region were heat pump households left cold more often, though the Northeast and Midwest samples are small. The survey counts only ducted heat pumps and measures cold homes, not equipment faults.
How often do apartment buildings lose heat?
In buildings with five or more apartments heated by a boiler, 10.4% of households, about 1 in 10, were uncomfortably cold for 24 hours or more in the winter of 2022-23 because the heating broke down, against 3.2% in buildings of that size heated by furnaces, according to RapidEye Research's analysis of the 2023 American Housing Survey. Renters with boilers reported 9.4%, against 2.7% for owners with boilers.
How many times does the heat break down in a winter?
Of the 3.4 million US households whose heating broke down and left them cold for a day or more in the winter of 2022-23, 46% had two or more breakdowns lasting six hours or longer and 22% had four or more. Among boiler households it was 57% with two or more, against 41% for furnace households, in RapidEye Research's analysis of the 2023 American Housing Survey.
Data sources
Every survey figure is our own computation from the Census Bureau's public file, run for the Department of Housing and Urban Development, and checked against the Bureau's published verification totals.
U.S. Census Bureau2023 housing survey public use microdatacensus.govU.S. Census BureauSurvey definitions, verification totals and variance methodcensus.gov