How Loud Is a Car? What the Pass-By Test Actually Measures

Every car sold in Europe has been through an acceleration pass-by test based on ISO 362, with microphones 7.5 metres from the centre of the lane. The car accelerates through a defined section under defined conditions and the highest A-weighted level is recorded.

The resulting limit is graded by power-to-mass ratio, on the reasonable premise that a 500 kW sports car and a small hatchback are not the same acoustic problem. Limits have also tightened in three phases since 2016.

The distinction matters more here than almost anywhere else on this site. 68 dB(A) is the highest a new ordinary car may measure under one specific accelerating manoeuvre at 7.5 metres. It is not what your car produces idling, not what it produces at motorway speed, and not what you hear on a pavement.

It is also why the EU and UN figures agree: EU Regulation 540/2014 incorporates the UN Regulation 51.03 limit structure and the same ISO 362-based method. That agreement is harmonisation, not independent confirmation.

For quiet electric and hybrid vehicles, regulators inverted the problem. At low speed these cars can be inaudible to pedestrians, so they must emit sound rather than avoid it.

The US standard, FMVSS 141, is instructive because it refuses to be a single loudness figure. For a stationary vehicle or one moving below 10 km/h with propulsion engaged, it specifies minimum A-weighted levels in thirteen one-third-octave bands, and requires at least four non-adjacent bands to meet them.

One is an overall A-weighted maximum during hard acceleration at 7.5 metres; the other is a per-band minimum at very low speed at a test position the table does not state. Putting 68 dB(A) and 41 dB in the same sentence as if they described the same axis would be meaningless.

The design reason for the band-by-band approach is audibility rather than loudness: a sound needs spectral content pedestrians can detect and locate against traffic background, which a single overall level cannot guarantee.

Cabin interior levels: we did not verify a published measurement of interior level at a stated speed from a source we could read in full, so this page does not quote one. Manufacturers publish such figures selectively and comparative test data was not accessible to us.

UN Regulation 138, the international rule for acoustic vehicle alerting systems, could not be read from the official document, so we have not quoted its minimum sound table. We used the US analogue instead and have labelled it as such rather than implying the two are the same.

Horns are regulated separately from vehicle pass-by noise, and this page does not cover them.

You cannot reproduce a type-approval pass-by test without a test track, a defined approach speed and instrument-grade equipment at 7.5 metres. Nothing consumer hardware produces is comparable with the figures in the first table.

What is worth measuring is your own situation: cabin level at a steady speed on the same stretch of road with the windows up, or how much level a passing car adds to your street. Our meter reports A-weighted digital signal level in dBFS when uncalibrated — relative, not sound pressure — and an estimated dBA figure after a one-time local calibration. Road and wind noise inside a moving car also loads the microphone with low-frequency energy, which is exactly where consumer microphone chains are least reliable.