What Hearing Conservation Actually Requires

Why this matters

Handing out earplugs is not a hearing conservation program, and it is not what the standard asks for first. 29 CFR 1910.95 hangs every obligation it contains off two numbers, both 8-hour time-weighted averages computed with a 5 dB exchange rate, and the two trigger different duties. Cross the lower one and you owe a program. Cross the higher one and you additionally owe an attempt at controls, and protectors stop being something you make available and become something that has to reduce the exposure to a stated level. Shops get this backwards in one specific way: they buy protection, skip the monitoring that would have told them which threshold they crossed, and end up with neither the control duty discharged nor the records that would show it.

The two thresholds, and how a level becomes one

The permissible exposure limit is an 8-hour time-weighted average of 90 dB A-weighted, from Table G-16 at 29 CFR 1910.95(a), with a 5 dB exchange rate: every 5 dB increase halves the permitted duration, so 95 dB A-weighted is permitted for 4 hours and 100 dB A-weighted for 2. The same paragraph sets a 115 dB A-weighted ceiling for continuous exposure and a limit of 140 dB peak sound pressure level for impulsive or impact noise, which is a peak figure rather than a time-weighted one.

The action level is an 8-hour time-weighted average of 85 dB A-weighted, equivalently a 50 percent dose, and it triggers the hearing conservation program at 29 CFR 1910.95(c).

Neither is a meter reading. Both are computed from the whole workday, by the method in Appendix A: for each segment, permitted duration is 8 divided by 2 raised to the power of the segment's level minus 90, over 5; dose is the sum of each segment's actual hours divided by its permitted hours; the equivalent 8-hour time-weighted average is 16.61 times the base-ten logarithm of the dose expressed as a fraction, plus 90. Levels from 80 dB A-weighted upward count toward the dose for the action level, and quieter segments contribute nothing.

What each threshold actually buys you in obligations

Read the last row twice. Protection is not a substitute for control above the permissible exposure limit. The standard asks for feasible controls and treats protectors as what covers the exposure while controls are pursued or where none is feasible.

Audiometry, the shift that matters, and where it becomes a recordable

The baseline audiogram is obtained within 6 months of an employee's first exposure at or above the action level, with a longer window where a mobile test service is used and protectors worn in the interim, and annual audiograms follow. All of that is 29 CFR 1910.95(g).

A standard threshold shift is an average change from that baseline of 10 dB or more at 2000, 3000 and 4000 Hz in either ear. It is measured against that employee's own baseline, which is why a missing or late baseline is expensive: without it there is nothing to measure a shift against for the rest of that person's employment.

A shift is a program trigger, and that is a separate question from whether the case is recordable. Under 29 CFR 1904.10 a work-related hearing loss case is recorded when there is a standard threshold shift and the employee's total hearing level is 25 dB or more above audiometric zero, averaged at 2000, 3000 and 4000 Hz in the same ear. Both conditions, not either. Treating the shift alone as recordable overstates the log; treating the 25 dB level alone as the test understates it.

The attenuation number, its basis, and the direction it is wrong in

The noise reduction rating on a hearing protector's package is there because of the EPA labelling requirement at 40 CFR Part 211 Subpart B, and it is a laboratory figure from subject testing under a stated method. Two things travel with it before it enters any arithmetic.

Its basis. The rating is derived against a C-weighted measure of the test signal. Where your exposure figure is A-weighted, which it is under this standard, OSHA's Appendix B method subtracts 7 dB from the rating before applying it. That 7 is a basis conversion, not a safety factor.

Its character. Real-world fit is worse than laboratory fit, consistently and in one direction, so the labelled rating is systematically optimistic rather than randomly scattered, which makes any allowance for it a one-sided bound rather than a plus-or-minus interval. OSHA's enforcement guidance for evaluating program effectiveness recommends halving the rating-minus-7 term for this reason. It is guidance on how to evaluate, not a number in the standard's text, and saying which is which matters if anybody quotes you.

Two employees, one gate, two different answers

Both work for the same shop. The gate is the same in both cases: compute the 8-hour time-weighted average from the whole workday using Appendix A, then place it against 85 and against 90.

Employee A, install technician. Personal dosimetry, A-weighted, slow response: 3 hours at 92 dB A-weighted during coring and demolition, 4 hours at 82 dB A-weighted on general install, 1 hour at 70 dB A-weighted driving and writing.

• Permitted duration at 92: 8 divided by 2 to the power 0.4, which is 6.06 hours. Dose contribution 3 over 6.06, which is 0.495.
• Permitted duration at 82: 8 divided by 2 to the power minus 1.6, which is 24.25 hours. Dose contribution 4 over 24.25, which is 0.165.
• The hour at 70 dB A-weighted contributes nothing, because the dose calculation counts from 80 dB A-weighted upward.
• Dose 0.660, or 66.0 percent. Time-weighted average is 16.61 times the logarithm of 0.660, plus 90, which is 87.0 dB A-weighted.

Outcome A. Above the action level, below the permissible exposure limit. The full program applies: monitoring, notification, audiometric testing, annual training, records, and protectors made available at no cost with a choice of types. No duty arises under 1910.95(b)(1), because Table G-16 is not exceeded. Protector use becomes mandatory for this employee if a shift appears.

Employee B, shop fabricator. Eight hours at a steady 93 dB A-weighted.

• Permitted duration at 93: 8 divided by 2 to the power 0.6, which is 5.28 hours. Dose is 8 over 5.28, which is 1.515, or 151.5 percent.
• Time-weighted average is 16.61 times the logarithm of 1.515, plus 90, which is 93.0 dB A-weighted, matching the steady level as it must for a single-segment full-shift exposure.

Outcome B. Above the permissible exposure limit. Everything employee A owes applies, plus the duty to institute feasible administrative or engineering controls, plus mandatory protector use attenuating to at or below 90 dB A-weighted.

The protector arithmetic for B, with basis and character as separate visible lines. A muff labelled 25 dB: the basis conversion takes 25 minus 7, which is 18; the one-sided field allowance halves that to 9.0 dB; estimated protected exposure is 93.0 minus 9.0, which is 84.0 dB A-weighted. That clears the 90 the standard requires and also the tighter 85 that would apply if this employee had a shift. Without the field allowance the same muff would have been credited with 93.0 minus 18, or 75.0 dB A-weighted, 9 dB more optimistic.

How little margin that hides. A protector labelled 15 dB gives 15 minus 7, which is 8, halved to 4.0, so 93.0 minus 4.0 is 89.0 dB A-weighted. That clears the 90 requirement by 1.0 dB and fails the 85 requirement outright. Same employee, same exposure, and the selection decides whether a future shift leaves the shop compliant.

Do not solve this by buying the highest number on the shelf. Over-attenuation in a moderate-noise area stops a worker hearing a backup alarm, a shouted warning or a change in the machine they are standing beside, so select for a protected level in the region the standard targets rather than for the largest rating available.

Sibling-rule check. Every level carries its weighting, its time basis, and whether it is a segment level or an 8-hour time-weighted average; no segment level is reported as a TWA or the reverse. The 5 dB exchange rate and the 80 dB A-weighted cutoff are the standard's own and are stated with it. The attenuation figure carries its basis and its character before entering any arithmetic, and the field allowance is applied as a one-sided bound rather than an interval. No dosimetry figure is background-corrected, because a personal dose is not a source-attribution measurement and the sibling procedure's margin gate belongs to that case rather than this one.

Where the general-industry Part is not the one that binds you

29 CFR 1910.95 is the general-industry standard. Where the work is construction, occupational noise exposure sits at 29 CFR 1926.52 with hearing protection at 29 CFR 1926.101, and those requirements are less prescriptive than the general-industry program. A field-service shop can fall under either depending on the job, so name which Part you are working to before quoting a requirement.

Separately, NIOSH recommends 85 dB A-weighted as an 8-hour time-weighted average using a 3 dB exchange rate, which is more protective than the OSHA figures above and is a recommendation rather than an enforceable requirement. Where a client's specification adopts the NIOSH basis, the same workday computes to a higher dose.

How to tell a real program from a filed one

Four checks, each of which fails independently.

A current exposure measurement for each job classification, repeated after the last significant change in equipment or process. Monitoring that predates a new machine describes a workplace that no longer exists.

A baseline audiogram for every employee in the program, dated inside the window after their first exposure at or above the action level. This is the record that cannot be reconstructed later.

Annual training on this workplace's own exposures, not a generic video. An employee who cannot name which of their tasks are the loud ones has not been trained on their job.

A protector on site whose attenuation, computed with the basis conversion and the field allowance, reaches the required level for the loudest classification. A cabinet full of a rating that does not close the gap is the most common quiet failure, and the arithmetic above catches it before anybody else does.

References

• 29 CFR 1910.95, occupational noise exposure in general industry, including Table G-16 at (a), the program trigger at (c), monitoring at (d), audiometric testing at (g), attenuation adequacy at (j), training at (k) and records at (m), plus Appendix A for the dose and time-weighted-average computation and Appendix B for estimating attenuation
• 29 CFR 1926.52 and 29 CFR 1926.101, occupational noise exposure and hearing protection in construction, where the work falls under that Part
• 29 CFR 1904.10, recording criteria for cases involving occupational hearing loss
• 40 CFR Part 211 Subpart B, the EPA labelling requirement that puts the noise reduction rating on the package
• See related: How to Work Out Whether a Noise Complaint Is Actionable; What a Weighting Network Is Doing to Your Reading