Workplace noise has a hard legal limit: 85 decibels over eight hours. Every 3 decibels louder halves that time.

An exposure standard is law, not guidance

The distinction matters more here than on most pages of this masthead. A code of practice is admissible evidence of what a duty holder should have known; an exposure standard is a regulation with a penalty attached. Regulation 56 defines the exposure standard for noise as an LAeq,8h of 85 dB(A), the eight-hour equivalent continuous A-weighted level, or an LC,peak of 140 dB(C), and regulation 57 makes a PCBU ensure the noise a worker is exposed to does not exceed it. The code is honest about what the number buys: the standard "protects most but not all people", research puts the level unlikely to cause hearing loss down at 75 dB(A) and peaks below 130 dB(C), so noise should be kept lower than the standard where reasonably practicable. Identifying whether you have a problem does not require a meter to start: the code's rule of thumb is that if you need to raise your voice to be heard by someone about one metre away, the noise is likely hazardous. A formal noise assessment by a competent person under AS/NZS 1269.1 is the next step where exposure is unclear, and the code notes it is not always a measurement exercise: a single machine with manufacturer noise data can settle the question on paper.

Three decibels doubles the dose

Decibels are logarithmic: a 3 dB increase doubles the sound energy, so every 3 dB of extra level reaches the same daily dose in half the time. The code's equivalent-exposure table is the clearest statement of what the standard means on a real shift:

Selected rows from Table 1 of the noise code: the time an unprotected worker can be exposed at each level before the LAeq,8h 85 dB(A) standard is exceeded, assuming the rest of the day is quiet. The full table runs to 130 dB(A), where the standard is exceeded in under one second.

The peak limit works differently. A single impact or explosive event above 140 dB(C), the code's examples are sledge-hammering and a gunshot, "can cause almost instant damage to hearing". No averaging, no dose arithmetic: one event is the breach. And because exposure is cumulative across a day, a worker who moves between several moderately noisy tasks can exceed the standard without any single task looking hazardous on its own.

The hierarchy starts at the purchase order

The code's control chapter runs the standard hierarchy, but its most practical page is about shopping. "Buy quiet" is its label for a purchasing and hiring policy that compares noise-emission data before plant is acquired, and this is backed by regulation, not aspiration: regulation 59 requires designers to design plant so its noise emission is as low as reasonably practicable, and requires designers, manufacturers, importers and suppliers to hand over noise-emission data with the plant. A duty holder is entitled to ask for that data and compare it across suppliers. The quieter-method examples are similarly concrete: bending metal in a press is quieter than hammering it, welding is quieter than riveting, gluing is quieter than nailing, and lowering materials beats dropping them. Isolation earns its own physics note: sound drops about 6 dB for each doubling of distance from the source, which is why moving a compressor often outperforms boxing it in. Even at levels that never threaten hearing, the code flags noise as a psychosocial hazard, suggesting levels below 50 dB(A) for work needing high concentration and below 70 dB(A) for routine work.

Earmuffs are the last resort, and they fail quietly

Personal hearing protectors sit at the bottom of the hierarchy for the same reason all PPE does: they control nothing at the source and depend entirely on behaviour. The code quantifies the failure mode. A protector delivering 30 dB of attenuation worn for a full eight-hour day delivers its full 30 dB; take it off for one hour of that day and the effective protection collapses to 9 dB. Where hearing protectors are the control for noise above the standard, a legal duty follows that many employers appear not to know exists. Regulation 58 requires the PCBU to provide audiometric testing for any worker frequently required to use hearing PPE for noise above the exposure standard: a baseline test within three months of starting the work, then at least every two years, conducted well into the shift so temporary threshold shifts are caught. The code adds the governance rules: consult workers before the program starts, give each worker their results with a written explanation, release individual results to anyone else only with the worker's consent, and treat any threshold shift or a reported tinnitus diagnosis as a trigger to review whether a control higher up the hierarchy is now reasonably practicable. It also recommends monitoring beyond the strict duty where hearing hazards stack: workers exposed to ototoxic substances or hand-arm vibration alongside noise from as low as 80 dB(A).

The injury the claims table cannot see

In Safe Work Australia's mechanism table for 2023-24, "sound and pressure" accounts for 200 serious claims nationally, 0.1 per cent of the total, with median time lost of 11.0 weeks and median compensation of $21,400. Read naively, that would make noise the safest hazard in Australia. The definition explains the illusion: a serious claim is one involving at least a working week off work, and noise-induced hearing loss arrives gradually, over years, without ever forcing a week's absence. Hearing loss claims are typically pursued as permanent-impairment claims, a category this table is not built to count. The 200 figure measures time lost, not hearing lost, and it is the reason this page leans on the exposure standard and the audiometric record rather than the claims data: for a slow-onset injury, the compliance paperwork is the early-warning system, not the compensation statistics. The same blind spot is why gradual-onset musculoskeletal disorders dominate the tables that noise never troubles.