An earplug labeled NRR 33 does not take 33 decibels off a worker's exposure. The Noise Reduction Rating is a laboratory number, and the mandatory Appendix B to 29 CFR 1910.95 already discounts it: subtract 7 dB before applying it to an A-weighted measurement. OSHA sometimes halves what is left, NIOSH cuts 25% to 70% off the label depending on the protector, and none of those three numbers are used for the same purpose.
What "derating" means, and why the label needs it
The NRR printed on the package comes from an EPA-regulated laboratory test - EPA requires manufacturers to put it on the packaging, and OSHA's Appendix B calls it "the most convenient method" for relating a protector to a worker's noise environment. In the lab, subjects are fitted under supervision, in quiet, with the protector seated correctly. On a jobsite, a foam plug gets rolled with dirty hands, half inserted, and pulled out to talk. The number does not transfer.
OSHA says so in Appendix B itself, in a note that sits above every method:
"The employer must remember that calculated attenuation values reflect realistic values only to the extent that the protectors are properly fitted and worn."
29 CFR 1910.95 Appendix B
Derating is the arithmetic that closes some of that gap. There are three separate schemes in circulation, and mixing them up is the most common mistake we see on hearing conservation paperwork.
| Scheme | Arithmetic | Status | Used for |
|---|---|---|---|
| Appendix B, 7 dB | NRR - 7, against an A-weighted level | Mandatory | Showing a protector meets 1910.95(j)(2) or (j)(3) |
| OSHA safety factor | (NRR - 7) / 2 | Enforcement practice | Deciding whether engineering controls are owed |
| NIOSH 1998 | NRR x 25%, 50% or 70% off, by protector type | Recommendation | Realistic selection; not an OSHA method |
The mandatory Appendix B methods
Appendix B is not guidance. The heading reads "This appendix is Mandatory," and 1910.95(j)(1) requires the employer to "use one of the evaluation methods described in appendix B." Which method you use is decided by your instrument, not by preference. The whole appendix comes down to one rule with two faces: C-weighted measurement, subtract the NRR; A-weighted measurement, subtract 7 dB from the NRR first.
| Instrument and weighting | Calculation |
|---|---|
| Dosimeter, C-weighted | C-weighted TWA - NRR |
| Dosimeter, A-weighted only | A-weighted TWA - (NRR - 7) |
| Sound level meter, A network | A-weighted TWA - (NRR - 7) |
| Sound level meter, C network | C-weighted average level - NRR |
| Area monitoring, A network | A-weighted area level - (NRR - 7) |
| Area monitoring, C network | C-weighted area level - NRR |
The result of every one of those lines is the same thing: the estimated A-weighted TWA under the ear protector. That is the number that has to clear the limits in the next section. Appendix B also lets an employer use one of three older NIOSH methods (#1B1, #1B2, #1B3) from HEW Publication No. 76-120 instead of the NRR; the appendix notes that the NRR method is itself "a simplification of NIOSH method #1B2." Almost nobody uses them, but they are legally available.
Run the numbers for your own crew with the NRR derating calculator, which applies all four methods side by side, and get the TWA it needs from the noise exposure TWA calculator.
Where the 7 dB comes from
The NRR is derived from C-weighted laboratory data, and C-weighting is nearly flat - it keeps low-frequency energy that A-weighting discards. Workplace exposures, meanwhile, are almost always measured in dBA, because the OSHA permissible exposure limit is written in dBA. Subtracting an NRR built on C-weighted data straight from a dBA number would over-credit the protector, so Appendix B applies a flat 7 dB correction to cover the spectral mismatch.
The (NRR - 7) / 2 safety factor: what it is really for
This is the most misquoted figure in hearing conservation. Plenty of training decks present (NRR - 7) / 2 as "the OSHA derating" for choosing earplugs. It is not. OSHA's letter of interpretation of December 16, 2021 states its enforcement policy in one paragraph:
"Under current enforcement policy, OSHA uses the unadjusted NRR to assess the adequacy of hearing protection when evaluating compliance with the hearing conservation program (HCP). However, in some circumstances, OSHA compliance personnel apply a safety factor of 50 percent to the laboratory-based NRR, above and beyond the 7 dB subtraction called for when using A-weighing measurements. OSHA uses this NRR-7/2 rating adjustment when considering whether an employer should implement engineering controls, but the safety factor adjustment is not used when evaluating the adequacy of HPDs under the HCP."
OSHA letter of interpretation, 12/16/2021
Read that carefully, because two things follow from it:
- For protector adequacy, the 50% factor does not apply. "Unadjusted" here means without the safety factor - the mandatory 7 dB subtraction in the A-weighted path still applies, because Appendix B says so.
- For engineering controls, it does. 1910.95(b)(1) requires feasible administrative or engineering controls when exposures exceed Table G-16. A compliance officer weighing whether an employer is leaning on earplugs instead of fixing the noise halves the lab NRR first - which is exactly the point of a safety factor.
(NRR - 7) / 2 on your own selection paperwork is conservative, defensible and often smart. Calling it "what OSHA requires" is wrong, and it will be corrected by anyone who has read the 2021 letter.The NIOSH derating, by protector type
NIOSH took a different route in its 1998 criteria document. Rather than one flat factor, it derates by how forgiving the protector is to bad fitting - because that is where the real-world loss lives. An earmuff is hard to put on wrong; a premolded plug is easy to leave sitting in the ear canal doing nothing.
| Protector | Subtract from label | Derated NRR | Estimated exposure (A-weighted) |
|---|---|---|---|
| Earmuffs | 25% | 22.5 | 100 - (22.5 - 7) = 84.5 dBA |
| Slow-recovery formable earplugs | 50% | 15.0 | 100 - (15.0 - 7) = 92.0 dBA |
| All other earplugs | 70% | 9.0 | 100 - (9.0 - 7) = 98.0 dBA |
Same label, same noise, and a 13.5 dB spread in the answer purely from which protector is in the box. That is the argument for the NIOSH scheme in a nutshell. It is a recommendation, not an OSHA method - NIOSH recommends, OSHA enforces - so use it to select protectors and to sanity-check the Appendix B result, not to replace it.
Double protection is not double the NRR
Plugs under muffs is the right call in very high noise, but the two ratings do not add. Sound reaches the cochlea through the skull as well as the ear canal, and that bone-conduction path puts a hard ceiling on total attenuation no matter what you hang on the outside of the head. OSHA's guidance is to add 5 dB to the NRR of the more protective of the two devices, then run the usual Appendix B arithmetic. An NRR 31 muff over an NRR 33 plug gives you a working NRR of 38, not 64.
What the protected number has to clear
Once you have the estimated A-weighted TWA under the protector, 1910.95(j) sets the bar in two lines:
- 90 dB, everyone. "Hearing protectors must attenuate employee exposure at least to an 8-hour time-weighted average of 90 decibels as required by paragraph (b) of this section" - 1910.95(j)(2).
- 85 dB after a standard threshold shift. "For employees who have experienced a standard threshold shift, hearing protectors must attenuate employee exposure to an 8-hour time-weighted average of 85 decibels or below" - 1910.95(j)(3).
- Re-evaluate when the noise changes. 1910.95(j)(4) requires the adequacy of attenuation to be re-evaluated whenever exposures rise enough that the protectors may no longer be adequate, and more effective protectors provided where necessary.
Note what is not in there: a requirement to over-protect. Burying a worker under 40 dB of attenuation in an 88 dBA area means they cannot hear a backup alarm, a shout, or the change in a bearing's pitch. Aim to land the protected exposure in the 75 to 85 dBA range, not as low as physically possible.
Why the label is a lab number, and what is changing
The rating on the box is governed by EPA, not OSHA, under 40 CFR part 211 subpart B - a regulation EPA issued in 1979 and has not replaced. On August 5, 2009 EPA proposed to update it (74 FR 39150), including adopting the ANSI/ASA S12.6-2008 Method A trained-subject fit protocol for passive protectors. As of this writing that proposal has not been finalized, so the NRR you read on a package today still comes from the older EPA test.
Two consequences worth knowing before someone sells you a solution:
- NRR (SF) is not an OSHA method. The subject-fit rating from the proposed EPA scheme "is not one of the methods listed in Appendix B when evaluating the attenuation of a HPD," per the 2021 interpretation. Using it instead of an Appendix B method can itself be a violation of (j)(1).
- A personal attenuation rating is not either. Asked whether a PAR from a fit-test system could replace the derating, OSHA pointed back to the same answer, and added that it "does not endorse or approve such equipment or products."
NIOSH has moved the other way. Its 2025 Science Policy Update (2025-104) recommends individual, quantitative fit testing to evaluate the attenuation a specific worker actually gets from a specific protector, superseding the 1998 derating scheme as best practice - the 1998 document had already said workers should ideally be fit tested, and only fell back on derating because no fit-test systems were commercially available at the time.
Bottom line
- Never subtract a raw NRR from a dBA level. A-weighted measurement means
NRR - 7; C-weighted means the NRR as printed. Appendix B is mandatory. (NRR - 7) / 2is OSHA's tool for deciding whether you owe engineering controls, not the test for whether a protector is adequate under the hearing conservation program.- The NIOSH 25 / 50 / 70% derating is the most realistic way to compare protector types before you buy. It is a recommendation, so document the Appendix B number too.
- Plugs plus muffs: add 5 dB to the higher NRR, not the two NRRs together.
- The protected TWA has to clear 90 dB, or 85 dB for anyone with a standard threshold shift - and does not need to go far below it.
- Derating estimates a population. Fit testing measures a person. NIOSH now recommends the second; OSHA still requires the paperwork from the first.