The 5 dB exchange rate is the whole difference
Two countries can measure the same jackhammer at the same distance for the same length of time and owe different duties, because they disagree about how noise trades against time.
OSHA halves the allowable duration for every 5 dB of extra level. Eight hours at 90 dBA, four at 95, two at 100, one at 105. That progression is Table G-16 in 29 CFR 1910.95, and Appendix A gives the formula behind it:
T = 8 / 2^((L - 90) / 5)Britain uses 3 dB, which is the equal-energy rule and the physically correct one: doubling sound energy is 10 log₁₀(2), or 3.01 dB. So the American table is more generous as levels climb, and the gap widens the louder the task. At 100 dBA it is two hours against roughly 48 minutes. Anybody who reaches for a British calculator to answer an American question gets a wrong number rather than a foreign one, which is the reason this page exists separately.
Dose, and why it beats an average
The dose adds each task's share of a full day. A task run for its full Table G-16 duration is 100% on its own, and partial exposures add up:
D = 100 x (C1/T1 + C2/T2 + ...)Then the dose converts to a single comparable figure, the 8-hour time-weighted average, using the constant written into Appendix A:
TWA = 16.61 x log10(D / 100) + 90That 16.61 is not a rounded convenience. It falls out of the 5 dB exchange rate itself, and it is why a doubled dose adds five decibels rather than three.
The practical consequence is that one loud task can decide a whole shift. Ninety minutes on a jackhammer at 105 dBA is already 150% of a dose before a five-hour afternoon at 90 dBA is counted. Averaging the two levels the way you would average temperatures gives a comfortable number and the wrong answer.
The two thresholds, and what each one buys
- 90 dBA TWA, a 100% dose: the permissible exposure limit. Over it, 1910.95(b)(1) asks for feasible engineering or administrative controls, with hearing protection used when those do not bring exposure within limits.
- 85 dBA TWA, a 50% dose: the action level. At or above it, a hearing conservation program is owed: audiometric testing, protection made available, training and records.
Note where the halves sit. Half the dose is 85 dBA, five decibels below the limit, not 45 dBA. Decibels are logarithmic and the intuition that halves the number is out by an enormous margin.
Check the part number before you rely on a figure
General industry is 1910.95. Construction noise is 1926.52, and it carries the same 90 dBA limit with the same durations under its own Table D-2. The difference is that 1910.95 sets out the hearing conservation program in detail and the construction standard does not, so a construction employer working to 1910.95 is generally doing more than 1926.52 spells out.
That distinction matters more than it sounds, because it runs the other way on confined spaces: there, the general industry standard expressly does not apply to construction, which has had its own rules under 1926 Subpart AA since August 2015. Two standards that look interchangeable can be interchangeable on one topic and not on another.
The button figures are typical, and yours will not be
It does not measure anything. The levels have to come from a sound level meter or a dosimeter used properly, and the typical figures on the buttons are there for a sanity check rather than as data about your site. It does not apply a hearing protector rating, because the thresholds are written against the exposure a worker stands in and the controls duty comes before the protection. And it does not tell you whether you are compliant. It gives you the dose and the average, and what follows from those is a judgement for whoever signs the program off.
Common questions
Why does this give a different answer to a UK noise calculator?
Because the two regimes trade noise against time differently. OSHA halves the allowable duration for every 5 dB, so 90 dBA is allowed for eight hours and 95 dBA for four. The UK Control of Noise at Work Regulations 2005 use 3 dB, which is the equal-energy rule that matches the physics of sound energy. At 100 dBA, Table G-16 allows two hours and an equal-energy calculation allows about 48 minutes. Same measurement, same worker, answers that differ by more than a factor of two, which is why this is a separate tool rather than a country setting on the British one.
What are the two thresholds?
The permissible exposure limit is 90 dBA as an 8-hour time-weighted average, which is a 100% dose. The action level is 85 dBA, a 50% dose, and reaching it is what triggers a hearing conservation program: audiometric testing, hearing protection made available, and training. Half the dose is five decibels below the limit rather than half the number, because decibels are logarithmic.
Does 1910.95 actually say "5 dB exchange rate"?
No, and it is worth being precise about that. The regulation never uses the phrase. What it publishes is Table G-16, pairing eight hours with 90 dBA, four with 95, two with 100 and one with 105, and the 5 dB relationship is what that table describes. Appendix A then gives the formula that reproduces it: T equals 8 divided by 2 to the power of L minus 90 over 5.
Does hearing protection bring the dose down on this calculator?
Not here, and that is deliberate. The dose this reports is the exposure a worker is standing in, which is the figure both thresholds are written against and the figure that decides whether a hearing conservation program is owed. 1910.95(b)(1) asks for feasible engineering and administrative controls first, with protection used when those do not bring exposure within limits. A calculator that quietly subtracts a rating off the box would hide the duty that comes before it.
Do these figures apply to construction as well as general industry?
Read the part number before relying on it. 1910.95 is the general industry standard, and construction noise sits under 1926.52, which shares the 90 dBA limit and the same Table D-2 durations. Where they differ is the detail of the hearing conservation program, which 1910.95 sets out at length and the construction standard does not, so a construction employer following 1910.95 is generally doing more than 1926.52 spells out rather than less.
Why can I not average the levels?
Because averaging decibels understates the loud parts. Two hours at 100 dBA and six at 85 averages to 88.75 dBA if you treat the numbers as ordinary arithmetic. The dose says otherwise: two hours at 100 is a full 100% on its own before the quieter six hours are counted at all. The loudest short task usually decides the day.
What happens to a quiet hour?
It counts as nothing, and the calculator shows it rather than hiding it. 1910.95(d)(1) measures the dose from 80 dBA upward, so anything quieter is outside the measurement rather than a small slice of it. A row you typed stays on screen contributing zero, because a row that vanished would look like a fault.
Is any of this the same as a noise survey?
No. This is arithmetic on levels somebody has already measured. The levels themselves have to come from a sound level meter or a dosimeter used properly, and the typical figures on the buttons here are starting points for a sanity check, not measurements of your site. Where a result lands near a threshold, that is the point to measure rather than to estimate.