2026 European heat-pump noise by brand and refrigerant: who is quietest in EPREL?

How quiet is the European heat-pump market in 2026?

The live EPREL heat-pump snapshot on 2026-06-18 contains 60,989 models from 777 manufacturers, with an overall average declared outdoor sound power of 61.3 dB(A) (market_index_snapshot / Househeating Pulse · Market Index v1, computed from EPREL Public API). Readers can inspect that broader universe in the live EPREL catalog and the current market index snapshot.

That average matters, but it can also mislead if taken as a buying rule. EPREL records declared sound power, not what a neighbour will hear at a bedroom window or what an installer will measure at a boundary fence. Sound pressure at a given distance depends on siting, reflections, defrost cycles, fan speed, casing design and operating mode. The EU EPREL registry itself is a product database, not a real-world acoustic survey.

Even so, the registry does show that very quiet entries exist. The 20 quietest products in the current dataset range from 1 dB(A) to 21 dB(A) by declared outdoor sound power (top_models / EPREL Public API via Househeating Pulse catalog). Those values are extraordinarily low, and some should be read with caution as declarations rather than field measurements. For model-by-model comparison, the most useful place to start is the live quietest leaderboard, then check each product page for context.

By product family, noise is not evenly distributed. Water-water units average 42.0 dB(A), the lowest declared outdoor sound power among types with reported data, versus 58.8 dB(A) for ground-water, 59.8 dB(A) for air-water and 64.1 dB(A) for air-air (type_efficiency / EPREL Public API · type aggregation). That spread of 22.1 dB(A) from water-water to air-air is much larger than the R290-versus-R32 gap discussed below, which already hints at the article’s main point: product type and lineup design appear to matter more than refrigerant alone.

The quietest EPREL models: the 20 lowest sound-power declarations

The table below lists the current 20 lowest sound-power declarations in EPREL as of 2026-06-18 (top_models / EPREL Public API via Househeating Pulse catalog).

For the narrower question of the 10 quietest model names, nine of the first ten are declared at 1 dB(A), with rank 10 also at 1 dB(A) (top_models / EPREL Public API via Househeating Pulse catalog). Those ten are BW 14 EVI, TBW 28 EVI, TWW 28 EVI, BW 11 EVI, TBW 22 EVI, TBW 50 EVI, TWW 110 WHR, TWW 48 EVI, HTW-MKT2-V500 and TWW 60 EVI (top_models / EPREL Public API via Househeating Pulse catalog).

The registry does not explain why some declarations are so low, nor does it provide installation geometry or test-lab notes in this corpus. That is why declared leaderboard position should be paired with the individual model pages and, where relevant, installer judgement.

Refrigerant counts inside this quietest set are also sparse. Among the top 20 quietest products, 2 use R290, 0 use R32, and 18 have no refrigerant recorded in this extract (top_models / EPREL Public API via Househeating Pulse catalog). That means R290 represents 10.0% of the top 20, R32 represents 0.0%, and 90.0% have refrigerant unrecorded here (top_models / EPREL Public API via Househeating Pulse catalog). So the registry slice does not support any strong refrigerant conclusion from the quietest-20 table alone.

Noise by heat-pump type: which product families are loudest and quietest?

Across EPREL product families, the quietest average belongs to water-water heat pumps at 42.0 dB(A) (type_efficiency / EPREL Public API · type aggregation). Next come ground-water heat pumps at 58.8 dB(A), then air-water heat pumps at 59.8 dB(A), while air-air heat pumps average 64.1 dB(A) and are the loudest group on average (type_efficiency / EPREL Public API · type aggregation).

Two caveats matter. First, the heat-pump water-heater category has 9,228 models, but this aggregation reports no average outdoor noise for that type (type_efficiency / EPREL Public API · type aggregation). Second, the very low water-water average comes from only 31 models, versus 30,452 air-water and 21,065 air-air models (type_efficiency / EPREL Public API · type aggregation). So water-water is quietest on average in this snapshot, but it is also a much smaller observed group.

R290 vs R32 and beyond: refrigerant patterns in declared noise

The corpus does not provide a direct average declared sound power by refrigerant group. So the registry does not answer, with numbers, what the average outdoor noise is for all R290 models versus all R32 models. It also does not support a measured claim that the R290 average is lower than the R32 average by a few dB(A), because that aggregation is absent from the dataset.

What the corpus does show is market presence. In the live snapshot, 13,935 models are declared as R32, while 537 are declared as R290 (market_index_snapshot / Househeating Pulse · Market Index v1, computed from EPREL Public API; refrigerant_universe / IPCC AR6 GWP table; EU Reg. 2024/573 phase-out schedule; EPREL declared codes). The same snapshot puts natural refrigerants at 3.27% of the market overall (market_index_snapshot / Househeating Pulse · Market Index v1, computed from EPREL Public API). Readers wanting the chemistry and policy context can use the refrigerants reference and the text of EU Regulation 2024/573.

The refrigerant reference table adds that R290 has GWP 0 and is classed A3, while R32 has GWP 771 and is classed A2L (refrigerant_universe / IPCC AR6 GWP table; EU Reg. 2024/573 phase-out schedule; EPREL declared codes). That may shape product engineering choices, but this corpus does not show refrigerant-by-noise averages. On current evidence, the stronger numerical story is that type spread is large and refrigerant-noise evidence here is incomplete.

Which brands dominate the market, and which brands are quietest on average?

EPREL’s brand concentration is high. Daikin Europe N.V. leads with 14,668 models and a 24.05% share, followed by Mitsubishi Electric Europe B.V. with 5,575 models and 9.14%, and JOHNSON CONTROLS HITACHI AIR CONDITIONING EUROPE SAS, SUCURSAL EN ESPAÑA with 5,207 models and 8.54% (brand_share / EPREL Public API · brand-share aggregation).

Among the 15 largest brands in the corpus, the lowest average SCOP belongs to Toshiba Carrier Europe S.A.S at 3.93, followed by Johnson Controls Hitachi at 4.18 and Panasonic Marketing Europe GmbH at 4.30 (brand_share / EPREL Public API · brand-share aggregation). The highest average SCOP among those top 15 belongs to Bosch Thermotechnik GmbH at 4.69, followed by GORENJE GOSPODINJSKI APARATI D.O.O. at 4.67 and Ariston SpA at 4.66 (brand_share / EPREL Public API · brand-share aggregation).

But the question of which large brands are quietest on average cannot be answered from this corpus. The brand-share table includes average SCOP, not average declared noise, and the brand-detail probe failed with “data unavailable” (brand_detail / (probe failed — data unavailable)). So the registry slice here records brand scale and average SCOP, but not brand-level average noise.

What buyers should make of EPREL noise declarations before choosing a model

Three practical conclusions follow from the data.

First, quiet models clearly exist across the market. The quietest 20 declarations run from 1 dB(A) to 21 dB(A) (top_models / EPREL Public API via Househeating Pulse catalog). Second, low-noise performance is uneven by type: water-water averages 42.0 dB(A), while air-air averages 64.1 dB(A) (type_efficiency / EPREL Public API · type aggregation). Third, this corpus does not show that refrigerant alone explains those differences, because refrigerant-level average noise is not reported.

For buyers comparing shortlisted products, the safe reading is that EPREL sound power is a useful screening metric, not a complete acoustic verdict. A lower declared sound power is generally better than a higher one, but installation still determines whether a unit is acceptable on site. Use the catalog filtered to air-water models or catalog filtered to R290 models, then compare class labels, capacities and SCOP. The methodology page explains how Househeating Pulse handles EPREL fields, and the HVAC glossary is useful if the distinction between sound power and sound pressure needs a quick refresher. Buyers can also pair acoustic filtering with the sizing calculator and payback calculator, since an oversized unit or the wrong operating strategy can undermine both comfort and economics.

Sources

• EPREL Public API via Househeating Pulse catalog — snapshot 2026-06-18
• EPREL Public API · type aggregation — snapshot 2026-06-18
• EPREL Public API · brand-share aggregation — snapshot 2026-06-18
• Househeating Pulse · Market Index v1, computed from EPREL Public API — snapshot 2026-06-18
• IPCC AR6 GWP table; EU Reg. 2024/573 phase-out schedule; EPREL declared codes — snapshot 2026-06-18
• (probe failed — data unavailable) — brand_detail probe, no as-of date returned

Continue reading

• How to read EPREL heat-pump data — A practical guide to labels, test points and common database pitfalls.
• R290 vs R32 heat pumps — What refrigerant choice changes, and what it does not.
• How to compare heat-pump noise properly — Sound power, sound pressure and why distance matters.
• How to choose the right heat-pump size — Capacity matching basics before price and brand comparisons.