How Loud Is an Airplane Takeoff? Certified Levels and What They Mean

Aircraft noise is regulated per aircraft type, not per event. Before a model can be sold, it is flown through a fixed procedure with microphones at defined reference points, and the resulting level goes on its type certificate. The point of that exercise is comparing designs under identical conditions — not telling a person on the ground what they will hear.

That regulatory metric is Effective Perceived Noise Level, EPNdB, which folds duration and tonal content into a single value. EPNdB numbers are not interchangeable with the dBA readings a sound level meter shows, and mixing the two is the single most common error in decibel charts.

The FAA does publish estimated maximum A-weighted levels measured under the same takeoff procedure, which is the closest thing to an apples-to-apples list. The spread across the fleet is enormous.

Read the table as a timeline and the story is clear. A 1970s 747 with JT9D engines sits near 100 dBA under the same procedure where a current-generation A320 sits at 73.3 dBA. That is not a small refinement; it is roughly a factor of several hundred in acoustic energy.

High-bypass turbofans are the main reason. Consequently, any decibel chart still quoting "jet takeoff = 140 dB" is describing neither a current aircraft nor a defined measurement position. We found no source supporting a figure like that for a civil takeoff at any documented listener position.

Airports run permanent and ad-hoc monitors, and their reports use a different metric again: CNEL, a 24-hour community noise exposure level with penalties for evening and night events.

San Francisco International's Millbrae ad-hoc report for April 2025, at a monitor about 3.9 miles from the runways handling roughly 1,217 aircraft noise events per day, gives an aircraft-attributed CNEL of 55 dBA against an average ambient of 73 dBA. In other words, at that location and over that period, non-aircraft noise dominated the 24-hour average even though individual overflights were plainly audible.

This is exactly why a single-event number and a community-average number should never be compared: they answer different questions, and the averaged one is designed to be lower.

If you want measured levels near an operating aircraft rather than a certification point, the occupational literature is where to look — with the caveat that it measures ground operations, taxi and auxiliary power units rather than the takeoff roll itself.

A NIOSH hazard evaluation at Frontier Airlines in Denver recorded 8-hour time-weighted averages of 83–89 dBA for mechanics, 83–86 dBA for ramp agents and 84–97 dBA for baggage handlers, with peak levels reaching 110 dBA near a Convair 580 turboprop. A later peer-reviewed NIOSH study of regional-jet ramp staff found 8-hour averages of 88–94 dBA for all six workers sampled, maximum slow-response levels of 111–116.5 dBA, and area measurements of 116–120 dBA at the tail near a running auxiliary power unit.

The figure most people actually want — how many decibels a departing airliner produces at, say, 500 metres from the end of the runway — is the one we could not find in any authoritative source. Certification data uses a fixed procedural geometry; community monitoring uses 24-hour averages; occupational data measures workers standing beside parked aircraft.

We are naming that gap rather than bridging it with arithmetic. Sound does not fall off from a moving, climbing, directional source the way a textbook point-source calculation assumes, and pretending otherwise would produce a confident number with nothing behind it.

You can measure what arrives at your position, which is a genuinely interesting question — but you cannot check it against any of the figures above, because none of them describe your position.

Our browser meter is deliberately explicit about this. Uncalibrated, it reports A-weighted digital signal level in dBFS: a relative number, not sound pressure. After a one-time local calibration against a trusted meter it reports an estimated dBA figure. Either way, an overflight is a rising-and-falling event, so a single instantaneous reading tells you far less than watching how the level changes as the aircraft passes.