Working at Height Risk Assessment Explained

A great many working at height risk assessments are not really assessments. They are lists. Hazard: fall from height. Control: use a harness, use a competent person, follow the method statement. Signed, filed, and silent on the one thing that matters, which is how anybody is going to reach the place the work has to happen and whether that place can hold them.

The message up front: the answer to a working at height risk assessment is usually a decision about the place and the means of access, not a list of precautions. A general risk assessment concludes with controls. A competent height assessment concludes with a chosen method of access and a recorded justification for it, plus who may do it, when it stops, and how somebody is recovered if it goes wrong. That is a materially different kind of output, and an assessment that does not produce it has not done the job.

1. Why height work needs its own assessment discipline

The general risk assessment method is sound and well covered elsewhere: identify the hazards, work out who could be harmed and how, evaluate the risk, decide on controls, record, review. For that process end to end read the complete guide to risk assessment, and for how the scoring works, the risk assessment matrix guide. Nothing here replaces either.

What makes height work different is the shape of the answer. For most hazards the controls sit alongside the task: you keep doing the task and add guarding, extraction, a permit, protective equipment. For height work the dominant control is a decision about where the person's body will be and what will be holding it up. Once you have chosen a tower rather than a ladder, or a mobile elevating work platform rather than a roof walk, or fixed edge protection rather than personal fall protection, you have made most of the safety decision.

That has three consequences. The assessment has to be done with knowledge of the actual place, because the place constrains the access options. It has to be done before equipment is committed, because otherwise it is reverse-engineered to justify the equipment already on the van. And its output is a decision with reasons attached, which is a different artefact from a control list and needs to be recorded as one.

The hazards themselves, the control hierarchy applied to falls, and the families of access equipment are the territory of the pillar article: working at height hazards, controls and requirements. This article stays on the assessment.

2. Where the duty comes from, by jurisdiction

Legal duties do not travel, and applying a foreign one is a common, avoidable error.

• Great Britain. The general duty to assess risk sits in the Management of Health and Safety at Work Regulations 1999 (SI 1999/3242), Regulation 3, under the Health and Safety at Work etc. Act 1974. Height work specifically is governed by the Work at Height Regulations 2005 (SI 2005/735), as amended by SI 2007/114. Access equipment is work equipment, so the Provision and Use of Work Equipment Regulations 1998 (SI 1998/2306) apply, and where lifting equipment is involved the Lifting Operations and Lifting Equipment Regulations 1998 (SI 1998/2307) apply. On construction projects the Construction (Design and Management) Regulations 2015 (SI 2015/51) add client, designer and principal contractor duties. Reportable incidents fall under the Reporting of Injuries, Diseases and Dangerous Occurrences Regulations 2013 (SI 2013/1471). Northern Ireland has separate instruments with different years and did not adopt the 2013 reporting set, so Great Britain citations are not automatically Northern Ireland citations.
• United States, federal. Fall protection in construction is 29 CFR 1926 Subpart M. In general industry the relevant body of rules is 29 CFR 1910 Subpart D, Walking-Working Surfaces, as rewritten by the rule effective 17 January 2017, with personal fall protection systems at 29 CFR 1910.140. State Plan states may impose different or stricter requirements. No federal standard requires a written task-step job safety analysis; that is guidance, and the closest legally required written hazard assessment in general industry is the written certification required by 29 CFR 1910.132(d) for protective equipment selection.
• United Arab Emirates. The binding law is Federal Decree-Law No. 33 of 2021 on the Regulation of Employment Relationships, administered by MOHRE, with occupational safety and health duties in Article 13. In Abu Dhabi the applicable framework is ADOSH-SF, the Abu Dhabi Occupational Safety and Health System Framework, Version 4.0, administered by the Abu Dhabi Public Health Centre. Say plainly what is true: UK HSE law and US OSHA regulations have no legal force in the UAE. They are voluntary benchmarks frequently written into specifications, but they are not the law there.
• Voluntary standards, everywhere. ISO 45001:2018, cited as amended by Amd 1:2024, is certifiable and requires the hierarchy of controls at clause 8.1.2, which is where the obligation to eliminate before you protect lives in a management-system sense. ANSI/ASSP Z10.0-2019 requires it at section 8.4 in the US. ISO 31000:2018 is guidance and not certifiable, so there is no accredited organisational certification to it. IEC 31010:2019 sets out risk assessment techniques, and note the designation: IEC 31010, not ISO 31010. None of these are law anywhere by themselves.

Where I have not named a height-specific regulation for a jurisdiction, that is deliberate: describe the duty in words and check the current instrument rather than quoting a number you half remember. For where this sits in the wider discipline, see what HSE actually covers.

3. The first question, and it must be first: can this be done without going up?

Every height assessment worth the paper begins by trying to make the height work unnecessary. Not as a box tick, as a genuine examination. This is the elimination tier of the hierarchy of controls, required by ISO 45001:2018 clause 8.1.2 and by ANSI/ASSP Z10.0-2019 section 8.4 in the US, and described by NIOSH on its public pages; for the hierarchy itself see the hierarchy of controls guide. Asking it properly means asking several sub-questions:

• Can the task be deleted? Some recurring high-level tasks exist because they were once specified and nobody has re-examined whether they are still needed, or whether they are being done at a frequency the asset does not warrant.
• Can the task be moved to the ground? Lowering devices, hinged or drop-down mountings, extension tools, and simply relocating the serviceable component are all ordinary engineering answers that remove the person from height entirely.
• Can it be done remotely? Camera inspection, drone survey, remote sensing and instrumentation can replace a visual check at height for some inspection tasks, though not for physical intervention.
• Can it be designed out? If something has to be serviced repeatedly at height, the honest answer is often to move the thing, not to keep sending people to it. That is a capital conversation, and it is frequently cheaper over the life of the asset than the access it replaces.
• If the height work is unavoidable, can the position be improved? Working from a stable platform inside guarding is different work from working at an unprotected edge, even at the same elevation. Changing the position is a real and often overlooked option.

An assessment that opens by selecting a ladder has skipped the only step that removes the hazard rather than managing it. That is the most common structural defect in these documents, and it is visible from the first line.

4. The place: what someone will stand on, and whether it will hold

This is the factor a generic risk assessment frequently fails to capture, and the one that kills people. For the actual location, the assessment has to establish:

• What the person will stand on or be supported by, and what that surface is made of.
• Its condition, and its capacity to carry the person plus tools, materials and any dynamic loading.
• Whether any part of it is fragile, and this includes surfaces adjacent to the work position as well as the work position itself, because people step sideways.
• Where the edges are, and where the openings are, including openings covered by something that is not a designed cover.
• What is overhead, including services, and what the clearances are as established on site by a competent person rather than assumed.

Here is the honest part that specifications tend to skip. On an existing building, the structural condition and load-bearing capacity of a roof, a walkway, a canopy, a false ceiling void or a plant platform is frequently unknown, and it is not established by looking at it. Age, water ingress, previous alterations, corrosion in hidden connections and coatings applied over degradation all make visual inspection an unreliable guide. Rooflights are a particular trap: painted, overpainted, replaced with a material resembling the surrounding deck, or simply dirty enough to read as solid.

5. The fall itself, and what is below

Two things need establishing. The first is the nature of the fall that is possible: from an edge, through a surface, into an opening, or below ground level into a pit, chamber, shaft or excavation. Falls below ground level are routinely missed because people read "height" as "up". They are not less dangerous for being downward, and where the space below is enclosed, confined space duties may also engage.

The second is the consequence, which depends heavily on what is underneath. People below, plant and machinery, vehicle traffic, water, further openings, hot or hazardous processes and stacked materials all change the outcome of the same fall. So does the landing surface.

The reminder that belongs in every one of these assessments: a modest height is frequently enough to kill, particularly onto a hard surface, onto an edge, or head first. The instinct to treat low-level work as low-risk work is wrong. I am not publishing a threshold here, because thresholds are jurisdictional and any figure I gave would be misapplied. The point stands without one: do not let elevation decide how seriously the task is assessed.

6. Duration and frequency, the factor most often mishandled

This is where height assessment practice is weakest, and the weakness runs in two directions at once.

Short tasks attract less planning and more improvisation. A job estimated at five minutes gets a ladder, a held-up hand, and a decision made on the spot, and no assessment, because the effort of assessing feels disproportionate. But exposure is not proportional to duration in any simple way. Most of the hazard is in the transition: getting up, getting positioned, getting the tool and the part up, getting back down. A short task contains the whole of that transition and skips the part where you are safely established on a platform.

Recurring short tasks accumulate far more exposure than a single long one, and are rarely assessed as a programme. A ten-minute filter change happening monthly at twelve locations is a very different risk profile from a single two-day roof repair, and it is the one that gets no assessment at all because each instance looks trivial. Aggregate the exposure across the year and the recurring task usually dominates, which makes it the strongest candidate in the building for engineered permanent access: a fixed platform, a fixed ladder with proper landings, a permanent guarded walkway, or relocation of the serviceable component.

7. Access, egress, the means of access, and anchorage

Getting to the place and getting back is a distinct part of the assessment, and where a meaningful share of falls actually happen. What has to be established:

• The route up and the route down, including whether the route down is the same one and whether it remains available once the work is under way.
• How equipment, tools, parts and waste get up and down. A person climbing with their hands full is the scenario that produces falls on otherwise well-controlled jobs. The assessment should conclude how materials move, not leave it to improvisation.
• The supporting conditions for the access equipment itself. Ground bearing, levelness, gradient, buried services, voids, basement slabs, drainage covers, soft standing after rain, and overhead obstructions and services on the approach as well as at the position.
• The erection and inspection status of the access equipment, established as a documented fact rather than assumed, and under the inspection regime that applies in the relevant jurisdiction. I am not stating intervals; take them from the applicable framework and the manufacturer.
• Whether the specific people are competent with that specific equipment. Competence is equipment-specific and configuration-specific. Being trained on one class of platform is not competence on another.

Anchorage deserves its own paragraph because it is where assumption is most dangerous. If the assessment is contemplating personal fall protection, it has to establish that a suitable anchor point exists, that it has been verified by someone qualified to verify it, and that it is reachable from where the person needs to be. Structural steel is not automatically an anchor. Pipework, cable tray, handrail, ductwork and roof plant are not anchors. Suitability is a specialist question involving the structure, the arrangement and the geometry of the position, and I am publishing no ratings and no clearance figures here precisely because those are the numbers people misremember and misapply. Get them from the competent person and the manufacturer for the actual arrangement. If the assessment cannot establish a verified anchor, personal fall protection is not an available control, and the assessment has to go back to the access decision rather than write in a harness and hope.

8. The people, the weather, and everyone else on site

The people. Competence for the task and the equipment, training that is current, physical fitness for the specific demands including those of an emergency, and whether anyone will be working alone. Lone height work changes the assessment substantially, because it changes who raises the alarm and how long a suspended or injured person waits. On multi-contractor sites, language and briefing are real risk factors: an assessment briefed in a language a crew member does not read has not been briefed to that crew member.

Environmental conditions and their variability. Wind is the obvious one, and it matters for platforms, towers, handling sheet materials and anyone at an edge. Also rain, ice, surface contamination, heat and the fatigue and dehydration that come with it, low light, and the fact that afternoon conditions are not morning conditions. The assessment needs a defined condition at which work stops, stated in terms the crew can actually apply on site, and a named person with the authority to call it.

Simultaneous operations. Other trades create hazards that are not in your task: openings cut or uncovered by somebody else, guarding temporarily removed, deliveries and lifting operations overhead, hot work nearby, scaffolding altered while you are on it, vehicle movements below. This is the coordination problem permits exist to manage, which is why height work around live plant frequently sits under one; see the permit to work guide. Where the work is a defined sequence, the step-level analysis in a job safety analysis is the right companion document.

9. Falling objects, exclusion, and recovery

Falling objects are part of the same assessment, not a separate one: tools, fixings, materials, debris, dislodged components and anything blown off the work position. The controls are tethering, containment, netting or covered access, and exclusion of the area below, which is usually the most reliable and the most frequently skipped because it inconveniences other people. Protective equipment for people below is the last tier, not the plan; see the guide to protective equipment for why it sits where it sits.

Recovery and emergency arrangements belong inside the assessment, and this is the omission I would look for first if I were auditing one. If the chosen arrangement is fall arrest, the assessment is not complete until it says how a suspended person will be brought down, by whom, with what, and how the alarm reaches them. The arrest is the beginning of the emergency, not the end of it, suspension after a fall is itself a medical emergency, and the people who will respond need to know in advance that they are the responders. The same applies to injury or illness on a platform, at an edge or on a roof: how does a casualty get down, and can the emergency services reach the position at all? On many buildings the honest answer changes the access decision.

10. The factors in one table

A consolidated view of what each factor requires and how it commonly goes wrong. This is a thinking aid, not a form to fill in.

11. An invented worked illustration

Below is a teaching example. Every line in it is invented by me to show how the factors resolve into an access decision. There is no real building, client, survey or crew behind it. Do not lift it. A real assessment is specific to the task, the structure, the equipment and the people involved, and the only way to produce one is on site with competent people.

Notice what the illustration does. It does not conclude "use a harness and be careful". It concludes with a chosen means of access, the reasons that choice was forced, an exclusion requirement, and a separate programme-level recommendation. On a different building, with a surveyed deck and verified anchors, the same task could legitimately resolve a completely different way, which is exactly why this cannot be a template.

12. The output, and getting it to the crew in time

A completed working at height risk assessment should leave behind, as a minimum:

• The chosen method of access, with a recorded justification for choosing it over the alternatives considered, including why elimination was not possible.
• The controls that go with that choice, including controls for the area below.
• Who may carry out the work, expressed as a competence requirement rather than a name where the crew may change.
• The stop conditions, and who has the authority to stop.
• The emergency and recovery arrangements.
• Any prerequisite that must be satisfied first, such as a structural assessment or an anchor verification, stated as a gate rather than a note.

Where the work needs a documented sequence too, the assessment pairs with a method statement, and the combined document is what the industry calls RAMS. See RAMS explained and what a method statement is. Keep the division clean: the assessment decides what will be done and why, the method statement says how.

13. Dynamic reassessment, and meaning it

Conditions on arrival routinely differ from conditions assumed. The hatch is blocked, the yard is occupied, the wind is up, the roof is wet, another trade has lifted a section of guarding, the fan is not where the drawing says. This is normal, not exceptional. So the assessment needs a defined trigger for reassessment and a named authority to stop, and the crew needs to know, credibly, that they may use it. That last clause is the one that fails. Where stopping is culturally expensive, crews adapt on the spot instead, and the adaptation is where the accident lives. A stop authority that exists on paper and is unusable in practice is worse than none, because it transfers the responsibility to the person least able to carry it. Practically: state the triggers in plain terms, name who to call, make the call routine rather than escalatory, and treat a stopped job as evidence the system worked.

14. Generic and task-specific assessments, honestly

Generic assessments get condemned in safety training and used everywhere in practice, which usually means the dogma is wrong rather than the practice. A generic assessment for a genuinely repeated task at a known location is legitimate and useful. It captures accumulated knowledge about a task done hundreds of times, it is more considered than anything a crew would write on the day, and it stops the organisation reinventing the same analysis weekly. The failure is not genericness; it is using a generic assessment for a task, or a place, that is not the one it was written for. The tell is simple: read it and ask whether it mentions anything identifiable about the actual location. If it would read identically for any building in the portfolio, it has not assessed a place, and for height work the place is most of the risk.

The workable pattern is a generic assessment for the task plus a short, recorded, site-specific confirmation on arrival that either confirms the assumptions or names the differences and triggers escalation. That keeps the accumulated task knowledge while restoring the location-specific judgement genericness strips out.

15. How height assessments fail

The failure patterns are consistent enough to list, and every one of them is visible on reading the document:

• Written from a desk. Nobody went and looked, so the place is described from a drawing, a photograph or a memory, and drawings of existing buildings are frequently wrong.
• The access method chosen first. The assessment is reverse-engineered to justify what was already on the van, which is why so many of them conclude "ladder" regardless of the task.
• Fragile surfaces not identified. Rooflights, degraded panels, temporary covers, and material that reads as deck from above.
• The structure assumed sound. Capacity treated as established because the surface looked solid and somebody walked on it last year.
• Recurring short tasks never assessed. Each instance too small to trigger paperwork, the aggregate never examined.
• Anchors assumed. A harness written into the controls with no verified anchor behind it, which is a control that does not exist.
• Recovery not addressed. Fall arrest specified with no plan for the arrested person.
• Conditions below ignored. The assessment protects the worker and says nothing about the occupied area, the walkway or the yard beneath.
• Not briefed to the crew. Filed, not communicated, or communicated in a language or format that did not land.
• Review date passed while the building changed. New plant, altered roof, new tenant, a route now blocked, and an assessment describing a place that no longer exists.

None of these are exotic. They are what happens when a height assessment is treated as a document to produce rather than a decision to make.

16. The facilities and maintenance angle

This is where my own work sits, and where the pattern is most fixable. A maintenance team's height exposure is not dominated by projects. It is dominated by a repeating list: filter changes, belt and bearing attention on roof plant, luminaire replacement in high spaces, gutter and drain clearance, facade cleaning, access to valves and dampers above ceilings, detector attention, camera and antenna work. Short tasks, high frequency, familiar locations, rarely assessed as a programme.

The argument I would make to any facilities director is to assess that list as a programme rather than reassessing the same ladder job forever. Pull the recurring high-level tasks out of the maintenance schedule, aggregate the annual exposure by location, then ask the elimination question at portfolio level: which of these positions justifies engineered permanent access, a relocated component, or a design change? That is a capital case built on real exposure data rather than on an incident, and usually a better case than people expect once the frequency is aggregated. The facilities maintenance management guide covers the programme view this sits inside, and work order types covers the classification that lets you pull recurring height work out as a set.

There is an asset-register point too. Permanent access installations, fixed ladders, walkways, edge protection, guardrail systems, anchor systems and rescue equipment are themselves assets. They degrade, they require examination, and they carry an examination history somebody has to hold. Treating them as building fabric rather than as registered assets is common, and it is how an anchor system quietly stops being a verified anchor system. Any competent maintenance system will hold this generically; the point is that the record exists and is retrievable when a crew asks whether the anchor on that roof is current. Where height work sits behind a permit, linking the permit to the work order and the asset keeps assessment, authorisation and equipment history in one place rather than three; see permit to work integration.

Final thoughts

The idea to walk away with is the one at the top. A working at height risk assessment is not a general assessment with falls added to the hazard list. Its job is to decide where a person's body will be and what will be holding it up, and to record why that was the right decision given what could be established about the place.

Which makes the quality test short. Does the assessment start by trying to make the height work unnecessary? Does it describe an actual identifiable place, including what is not known about it? Does it conclude with a chosen means of access and reasons? Does it say when to stop and who may stop it? Does it say how somebody comes down if it goes wrong? And did it reach the crew before they mobilised? An assessment that answers those six is doing real work. One that does not is a filing exercise, however thorough the hazard list looks.

And a last word about the small jobs. The assessments that get the most attention are for the big, visible, obviously dangerous work, and those are usually done best. The exposure that actually accumulates in a maintenance operation is in the ten-minute task nobody thought warranted a document, repeated across a portfolio, for years. Assess that properly, aggregate it, and engineer the access out where the numbers justify it. It is sitting in the planned maintenance schedule waiting to be counted.

For the hazards and the equipment families themselves, go back to the pillar: working at height hazards, controls and requirements. Primary sources worth going to directly: HSE (Great Britain) , OSHA (United States) and ISO .

Related reading: Working at height: hazards, controls and requirements, Risk assessment: the complete guide, Risk assessment matrix, Hierarchy of controls, Job safety analysis, RAMS explained, What is a method statement, Permit to work, Protective equipment, What is HSE, Facilities maintenance management, Work order types, Permit to work integration.