How to conduct a job hazard analysis

What a JHA is for

A job hazard analysis breaks a task into its steps, identifies what could hurt someone at each step, and records the control that addresses it. The output is a document, but the value is in the conversation that produces it — a crew working through their own job in sequence will surface things no checklist anticipates.

OSHA publishes a well-regarded booklet on the technique, Job Hazard Analysis (OSHA 3071), which is worth reading in full. It describes an approach rather than a mandated form. There is no single legally required JHA format, and any vendor telling you otherwise is selling a template.

Step 1 — Choose the job

You cannot analyze everything at once, so priority matters. Jobs worth analyzing first:

• Those with a history of incidents or near misses — you already have evidence something is wrong.
• Those where the realistic worst case is severe, even if nothing has happened yet.
• New jobs, or jobs where the method, equipment or materials have recently changed.
• Non-routine work — shutdowns, maintenance, first-time tasks — where habit provides no protection.
• Jobs complex enough that written instructions already exist and are not being followed.

Scope the job narrowly enough to be analysable. "Warehouse operations" is not a job. "Replacing a damaged guardrail section on the mezzanine" is.

Step 2 — Involve the people who do it

A JHA written at a desk describes the job as management imagines it. The people doing the work know the shortcuts, the awkward reach, the step everyone skips because the official method takes twice as long. That gap between work-as-imagined and work-as-done is where injuries live.

Watch the job being performed if you can, and ask the crew to walk you through it in their own words. If they describe a step that is not in the official procedure, that step goes in the analysis — you are documenting the job as it is actually done, then deciding whether it should change.

Step 3 — Break the job into steps

List the steps in the order they occur. Each step should describe an action, not a detailed technique. A useful test: most jobs land somewhere between about five and fifteen steps.

• Too coarse: "Replace the guardrail." That is the job, not a step.
• Too fine: "Pick up the wrench. Position the wrench on the bolt." You are writing a manual, not an analysis.
• About right: "Set up the exclusion zone below." "Remove the damaged section." "Position and secure the replacement."

Do not forget the beginning and the end. Setup, access to the work area, cleanup and return to service are where a surprising share of injuries occur, and they are the steps most often omitted.

Step 4 — Identify the hazard at each step

Take each step and ask what could go wrong, being specific about the mechanism. "Unsafe conditions" is not a hazard; "fall from height while leaning past the open edge" is. Prompts that help:

• Gravity — falls from height, falling objects, loads that can shift or roll.
• Stored energy — electrical, hydraulic, pneumatic, thermal, springs, suspended loads.
• Motion — vehicles, moving machinery, pinch and crush points, struck-by.
• Substances — chemical exposure, dusts, fumes, oxygen deficiency in confined spaces.
• Ergonomics — awkward posture, repetitive loading, manual handling.
• Environment — heat, cold, noise, poor light, weather, uneven ground.

Ask what happens if a step is done out of order, done by someone unfamiliar with it, or interrupted halfway. Record the realistic consequence, not just the hazard name — "fall of roughly four metres onto concrete" is more useful to a reviewer than "fall hazard", and it drives the risk rating honestly.

Step 5 — Choose controls

Work down the hierarchy of controls, in order, and only move to the next level when the one above is genuinely not practicable. The order reflects reliability: the earlier controls remove the hazard, while the later ones depend on human behavior every single time.

1. Elimination — remove the hazard. Can the work be done at ground level instead?
2. Substitution — a less hazardous material, method or piece of equipment.
3. Engineering controls — guarding, ventilation, interlocks, permanent edge protection. They work whether or not anyone is paying attention.
4. Administrative controls — procedures, permits, training, rotation, exclusion zones. They depend on people following them.
5. PPE — the last line, protecting one person if everything else fails.

Two rules of thumb. A control that requires perfect compliance under time pressure is weaker than it looks on paper. And if the only control for a serious hazard is "be careful" or "trained personnel only", the analysis is not finished.

Step 6 — Get PPE in the right place

PPE belongs in a JHA, but at the step where it is needed rather than as a blanket list at the top. "Hard hat, hi-vis, gloves, steel toes" applied to the whole job tells a worker nothing — they already knew that. "Fall arrest harness anchored above the working position, for steps 3 and 4 only" is actionable.

Where PPE is the primary control for a serious hazard, say so explicitly. It flags to a reviewer that the higher levels of the hierarchy were not achievable, and invites the question of whether that is really true.

Step 7 — Document and sign off

The written analysis should let someone who was not in the room understand the job. At minimum: the job and location, the date, the person responsible, the steps, the hazards, the risk rating, the controls, and the required PPE — plus a signature from whoever reviewed and approved it.

Then make it reachable at the point of work. A JHA in a binder in the site office is a compliance artefact; a JHA the crew opens before starting is a control. Brief it as part of the pre-task discussion and let people challenge it — the step somebody argues with is usually the one that was wrong.

When to review it again

A JHA is not a one-time document. Trigger a review when:

• The method, tools, equipment or materials change.
• An incident or near miss occurs on that job — the analysis missed something, or a control failed.
• New people begin performing the job, or the crew composition changes significantly.
• The environment changes — a different site, season, or adjacent work.
• Nothing has happened for a long while. A periodic review catches drift between the written method and the practised one.

Common failure modes

• Written once, filed forever. An analysis nobody has opened in three years describes a job that no longer exists.
• Copied between jobs. A JHA reused for a superficially similar task inherits the wrong hazards and misses the real ones.
• Every risk rated medium. If nothing is ever high, the rating carries no information and reviewers stop reading it.
• Controls that name a document. "Follow SOP-114" is a reference, not a control. State what the person actually does.
• Written for the auditor. The analysis is complete, signed, and describes an idealised job nobody performs. This is the most common failure and the hardest to detect from the paperwork.

Sources

• OSHA — Job Hazard Analysis (OSHA 3071)
• OSHA — Hazard identification and assessment