What Is Risk in Health and Safety? The Definition, the Formula, the Rating

The definition, and why it has two halves

In health and safety, risk is the possibility that a hazard causes harm — and the possibility is a product, not a feeling. The two halves are the likelihood (how probable is it that the hazard does the harm) and the consequence (if it does, how bad is it). Multiply the halves and you get the risk, and the number is what the management review reads, the auditor checks, and the control order is built from.

The distinction from the hazard is the whole point. The hazard is the thing that can do the harm; the risk is the bet on whether it does, and on how hard. A locked, ventilated solvent store and an open drum on a busy bench are the same hazard — the solvent — and two very different risks. The conditions moved the likelihood and the consequence; the hazard did not move at all.

Hold the definition as a question: how likely is it that this hazard harms a person, and how bad would the harm be? The answer, in numbers, is the risk.

The formula: likelihood times consequence

The working formula is the product of the two halves, each scored on a 1-to-5 scale. The likelihood scale: 1 rare, 2 unlikely, 3 possible, 4 likely, 5 almost certain. The consequence scale: 1 negligible, 2 minor (first aid), 3 serious (medical treatment, lost time), 4 major (multiple injuries), 5 catastrophic (fatality).

Multiply and you get the risk rating, 1 to 25, on the 5×5 matrix. The standard bands: 1 to 4 is low — acceptable, maintain the controls and monitor. 5 to 9 is medium — tolerable only if the controls are in place, and the work goes ahead to an acceptable level of residual risk, the ALARP band. 10 to 15 is high — intolerable without further control; the action is priority, with the owner and the date. 16 to 25 is very high — the work stops, or is restricted, until the control is in place.

The matrix is a tool, not a law. The UK system expects the assessment to be suitable and sufficient — the scales and the bands are the method, and the competent person owns the call where the score is close. But the score is the shared language: the crew, the supervisor and the management review all read the same number, and the argument is about the control, not about the feeling.

ALARP: the band that does the work

The middle band is where the real decisions live. ALARP — As Low As Reasonably Practicable — is the rule that the residual risk in the medium band must be brought down as far as is reasonably practicable, and reasonably practicable has a test: the cost of the control is not grossly disproportionate to the benefit of the risk reduction.

A guard rail on a mezzanine edge is not grossly disproportionate to the fall it prevents; the ALARP argument does not survive the arithmetic. A full automation line for a risk of 6 with a first-aid consequence is the case where the disproportionality argument has force — the cost is grossly disproportionate to the benefit, and the residual is accepted with the control that is in place.

ALARP is the middle of the matrix made into a duty. The low band is the monitor, the high band is the priority action, and the medium band is the ALARP argument — the place where the competent person decides, in the open, that the cost of the next control is grossly disproportionate to the benefit. That decision, written down, is the record the investigation reads.

Risk vs the words around it

Risk sits inside a family of words the interview keeps mixing. The hazard is the source; the risk is the likelihood and severity; the danger is the state of being exposed; the harm is the result that actually happens. And risk assessment is the whole process — identify the hazards, evaluate the risks, record the findings, review them — while risk control is the hierarchy that acts on the score: eliminate, substitute, engineer, manage, and PPE as the last line.

A residual risk is the risk that remains after the controls — the post-control score on the matrix, and the one the review reads. A tolerable risk is the residual that the ALARP test has accepted; an intolerable risk is the one that the work stops for. The vocabulary is small, and it is the same vocabulary in the UK, the US and the ISO system — the MHSWR, the OSHA general duty and the ISO 45001 clause 6.1 all work from the hazard, the risk, the control and the residual.

The chain, in one line: the hazard is the source, the risk is the likelihood and severity, the control moves the risk, the residual is what is left, and the ALARP test decides whether the residual is accepted.

Worked example: scoring the workshop

The same workshop, scored on the matrix, and the control order the score produces.

The pre-control scores set the order: the press first (the 20), the mezzanine and the forklift next (the 15 and the 12), the handling after (the 9), the slip last (the 4). The post-control scores are the residual — the ALARP argument, written down, per line. The matrix did the two jobs: it ordered the work, and it recorded the residual that the review reads.

The mistakes that score the risk wrong

The risk that is scored wrong is scored wrong for a small set of repeatable reasons. The consequence is understated: the fall is scored as 3 when the height makes it a 5 — the consequence scale is the one the assessor shrinks, and the high line that should be a very high comes out a high. The likelihood is a feeling: the 3 that is really a 4, because the task is done every shift, not once a month. The control is scored before it is in place: the post-control is written for the guard that is on order, not for the open edge that is there now.

The matrix is changed to fit the answer: the band is moved, or the scale is re-labelled, to make the residual acceptable — the ALARP decision becomes the ALARP assumption, and the investigation reads the change. The residual is not written: the pre-control is scored and the control is named, but the post-control is never put down, and the review has no residual to check.

The score is the shared language of the site, and the language is only as good as the discipline behind it. The consequence from the height, the likelihood from the frequency, the post-control from the control that is actually in place — that is the scoring, and it is the start of the control order.