NIOSH vs OSHA Noise Limits

Two American noise limits get quoted, they are not interchangeable, and almost every argument about workplace noise comes down to which one someone has in mind.

NIOSH recommends 85 dBA averaged over eight hours, on a 3 dB exchange rateNIOSH. OSHA enforces 90 dBA over eight hours, on a 5 dB exchange rateOSHA.

The first is what the research supports. The second is the law in a US workplace. They are five decibels and one exchange rate apart, and that is a much larger gap than it looks.

What the gap actually costs

This is the number that ends the debate, and it does not come from either agency — neither publishes the other's risk estimate, which is precisely why the comparison needs a third source.

A forty-year working life at the NIOSH limit still carries a risk of noise-induced hearing loss of about 8%. The same forty years at the OSHA limit carries a risk of about 25%Sheppard et al..

One in four against one in twelve, for the same career, decided by which document your employer is held to rather than by anything about your ears.

Why the exchange rate matters more than the level

The five-decibel difference in the limit is the visible half. The exchange rate is the half that compounds.

NIOSH uses 3 dB because 3 dB is a doubling of sound energy — the equal-energy principle, where the allowed time halves for every 3 dB increaseNIOSH. OSHA uses 5 dB, so under its rule the allowed time halves only every 5 dB.

Follow both up from their own starting points and they diverge fast. NIOSH allows four hours at 88 dBA and one hour at 94 dBANIOSH. OSHA, starting five decibels higher and halving more slowly, permits far longer at the same levels — the two rules are not a small correction apart at the loud end, they are a different shape.

The effect of that shape has been estimated. Some studies indicate that adopting a 3 dB exchange rate in OSHA regulations would identify one and a half to three times as many workers as being potentially at risk of noise-induced hearing lossSheppard et al.. That is the population the difference between the two rules is currently sorting into "protected" and "not counted".

What the two agencies agree on

More than the argument suggests, and the agreements are what make the comparison possible at all.

Both are written in A-weighted decibels, so the numbers are the same kind of number and can at least be set side by side. Both are eight-hour time-weighted averages rather than instantaneous levels, which means neither can be checked by glancing at a readout. And both are built on the idea that exposure is level multiplied by time — they disagree about the multiplier, not about the principle.

NIOSH's recommended meter settings make the shared vocabulary explicit: A frequency weighting, Slow response, and its own 3 dB exchange rateNIOSH. Change the weighting and neither limit means anything; change the exchange rate and you have moved from one agency's model to the other's without saying so.

That last point is where most confusion starts, and the next section is why.

Why OSHA's number is higher

Not because anyone thinks 90 dBA is safer than 85. Because an enforceable standard and a health recommendation are answering different questions.

A recommendation says what the evidence supports. An enforceable standard has to be defensible as achievable at the time it is written, across every industry it covers, which brings feasibility and cost into a calculation that a recommendation never has to make. NIOSH itself notes that OSHA, in accepting a NIOSH recommendation as the basis for a mandatory standard, is required to consider cost and feasibility under the Occupational Safety and Health Act of 1970NIOSH.

That is a legitimate reason for the two numbers to differ. It is not a reason to treat the higher one as a safety threshold, and the risk figures above are what the difference buys.

Which one applies to you

If you are an employee in a US workplace, the enforceable limit is OSHA's, and that is what your employer's programme is built around. The NIOSH figure is the one to know if you want to understand what the legal minimum leaves on the table.

If you are anywhere else — at home, at a gig, in a workshop of your own — neither applies, and the more useful thing to carry is NIOSH's simpler rule: whenever levels exceed 85 dB(A), take precautions, regardless of how long the noise lastsNIOSH. The dose table is for shifts. That sentence is for everyone.

If you are trying to work out whether a day was a problem, neither limit answers it on its own, because both are about averages over time rather than levels at a moment. The noise dose calculator does that arithmetic on the NIOSH model, which is the one this site uses throughout.

What this site does with the disagreement

It publishes NIOSH's model and says so.

Every safe-time figure, every dose percentage, and the 3 dB rate behind them come from NIOSH rather than OSHA. That is a deliberate choice, and the reason is that the 3 dB rate follows the physics: sound energy really does double every 3 dB, and a 5 dB rate is a policy compromise rather than a description of what happens to your ears.

The consequence is worth stating plainly. Nothing on this site is an OSHA compliance determination, and a reading that would clear the OSHA limit can still be over the NIOSH one. If you need to know whether a workplace complies with anything, you need a hearing conservation programme and instrument hardware, not a browser page.

Where the argument usually goes wrong

Three mistakes account for most of it.

Quoting a level without a duration. Both limits are eight-hour averages. "The machine reads 92" is not a comparison with either of them until you know how long anyone stands next to it. That is what 85 decibels works through in detail.

Mixing the two rules. Taking OSHA's 90 dBA limit and applying NIOSH's 3 dB exchange rate produces a rule neither agency publishes, and it turns up constantly in noise calculators. Pick a model and stay in it.

Treating the legal limit as a safety line. It is a floor for enforcement. The gap between "lawful" and "unlikely to damage hearing" is the whole subject of this page, and it is one in four against one in twelveSheppard et al..

If the noise is not yours

Neither limit is written for a flat, a street, or a neighbour, and reaching for occupational figures in a domestic dispute is a reliable way to lose it. The thresholds that matter there are much lower and the evidence that matters is a record rather than a reading, which is what document neighbor noise is about.

For the room you are in right now, the online decibel meter on this site gives an uncalibrated estimate in dBA — the same weighting both limits are written in, which at least means the number is the same kind of number, even if it is not one either agency would accept.