Safety Signs: Types, Symbols and Colours Explained

Walk into almost any plant room, loading bay or substation corridor and you will find signs. Some of them are current, some of them refer to equipment that was removed years ago, a few contradict each other, and a handful are so surrounded by other signs that nobody has read any of them in a decade. Signage is cheap, visible and easy to specify, which is exactly why organisations reach for it, and exactly why it accumulates faster than it is ever reviewed. The useful way to think about safety signs is not as decoration or as compliance wallpaper, but as a communication system with a specific job, real limits, and a maintenance requirement of its own.

The message up front: a safety sign is an administrative control. It informs, warns or instructs, and it removes no hazard whatsoever. It works only if someone sees it, understands it, believes it and is able to act on it, and each of those four conditions fails routinely. So the first question about any sign is what it is doing there instead of a control. The second question, almost never asked, is whether the sign next to it is making it unreadable.

1. A sign is an administrative control, and it removes nothing

Start with the placement of signage in the hierarchy of controls, because everything else follows from it. The hierarchy is a principle rather than a standard in its own right: it is required by ISO 45001:2018 (as amended by Amd 1:2024) at clause 8.1.2, which is an international voluntary management-system standard and law nowhere by itself; it is also required by ANSI/ASSP Z10.0-2019, a US voluntary consensus standard, at section 8.4; and it is described by NIOSH in the United States, which is a research agency with no regulatory power. US OSHA expresses a preference for engineering and administrative controls over personal protective equipment in places, but it does not codify the hierarchy of controls as a term in regulation. If you want the full treatment, the hierarchy of controls guide sets out the tiers properly.

Signage sits in the administrative tier, near the bottom. That is a statement about mechanism, not a slur. Elimination and substitution remove the hazard; engineering controls, and the reorganisation of work that ISO groups with them, keep people away from it whether or not anyone is paying attention; administrative controls depend entirely on human behaviour going right. Personal protective equipment sits below signage again, and is covered in the PPE guide.

The chain a sign has to complete is long. The viewer must be looking in the right direction at the right moment, and the sign must be legible from where they are in the light that is actually there. They must recognise what it means, which is not the same as being able to see it. They must believe it applies to them now, and not have learned that most signs on this site are stale. And they must be in a position to comply, which fails immediately if the sign says to wear hearing protection and the dispenser has been empty for a fortnight. Four soft conditions, none guaranteed. That is what an administrative control is.

Having said that, signage is necessary, and the fashionable dismissal of it is as wrong as the over-reliance. Residual hazards exist after every reasonable control has been applied, and people are entitled to know about them. Contractors, visitors, delivery drivers and new starters arrive with no site knowledge at all, and signage is the only information channel that reaches them passively. And in an emergency, when the lights are out, the alarm is sounding and nobody is reading a procedure, a sign may be the only instruction available. Emergency escape signage is worth calling out specifically: it is not a weak control substituting for something better, it is the right control for its job, and there is no engineering control that tells a stranger which way the exit is. Judge each sign by whether it is doing its own job or standing in for a control that was never installed.

2. The message categories, which are the real structure of the subject

The genuinely transferable content in safety signage is the set of message categories. Sign systems around the world organise signage into a small number of message types, and although the graphics differ, the categories are broadly recognisable across systems because they reflect what there is to say rather than how a committee chose to draw it.

Two categories are worth distinguishing carefully because they are constantly muddled. Warning signage points at a hazard; safe-condition signage points at safety. They are opposite in intent, and a viewer who confuses the two in an emergency is in real trouble. A related distinction: fire-equipment signage tells you where the extinguisher is, which is not the same as escape signage telling you where the exit is, and it is not the same as the detection and alarm hardware that warns you in the first place. For that side of the subject see the fire alarm system guide, and for the statutory-inspection side of life-safety assets the elevator, fire and life-safety PM guide.

3. How shape and colour encode the message

Here is the design principle that makes signage work at all, and it is common across the major systems even though the specific implementations differ. Each message category is given its own combination of shape and colour, consistently, so that the viewer reads the kind of message before reading any text or recognising any symbol. The shape and colour carry the grammar. The pictogram and any wording carry the detail.

That ordering matters in real conditions. A person walking quickly down a corridor, at an angle, in poor light, possibly with no fluency in the site's written language, will resolve a coloured shape further away than a symbol, and a symbol further away than text. If the encoding is consistent across the site, they get a usable first-order message, this is something I must not do, or this is a hazard, or this is where safety is, before they are close enough to read the specifics. That is why consistency within a site matters more than the elegance of any individual sign.

What I am deliberately not doing here is publishing a colour code. Which colour means which category, which shape carries which message, how large a sign must be, and how far away it must remain legible are all set by the standard adopted in your jurisdiction, and they are not the same between systems. Reproducing one system's assignments as though they were universal is the single most common failing in articles on this subject, and it is actively dangerous for a multinational reader. Specify from the applicable standard, not from a web page. Where this article refers to a category's typical appearance, treat it as an illustration of the principle and nothing more.

4. The standards genuinely differ, and that is substance, not a caveat

There is no single global safety-sign standard, and the differences between the major systems are not cosmetic. Two families dominate, and they were built on different philosophies.

ISO 7010:2019, "Graphical symbols, safety colours and safety signs, registered safety signs", is an International Organization for Standardization document, voluntary and law nowhere by itself. It has been heavily amended and is currently under revision, so cite it as ISO 7010:2019 as amended and check the current published status before specifying. Its approach is registered pictograms that are symbol-only and wordless by design, so that the sign is language independent. The European adoption is EN ISO 7010, whose amendment sequence has historically lagged the ISO one, which means an EN-marked sign and an ISO-marked sign are not automatically at the same amendment level. Alongside it sit the design-principle documents: ISO 3864-1:2011 for design principles, ISO 3864-2:2016 for product safety labels, and ISO 3864-4:2011 for the colorimetric and photometric properties. All are ISO, all are voluntary, all are paywalled, and none of them is law anywhere on its own.

The ANSI Z535 family in the United States takes a different route. It is published by ANSI with NEMA as secretariat, it is a US voluntary consensus standard family and not law by itself, and it is built around a signal-word panel plus text with an optional symbol, which makes it fundamentally text-dependent rather than language independent. The parts that matter to a facilities reader include ANSI Z535.1-2022 on safety colours, ANSI Z535.2-2023 on environmental and facility safety signs, ANSI Z535.3-2022 on criteria for safety symbols, ANSI Z535.5-2022 on safety tags and barricade tapes, ANSI Z535.6-2023 on safety information in product manuals, and ANSI Z535.7-2024. In broad practice, ISO 7010 and EN ISO 7010 are the reference in the European Union, the United Kingdom, much of Asia and the Gulf, while ANSI Z535 is the reference in the United States. That is a description of practice, not something any standard states, and it does not tell you what your own jurisdiction's law requires.

The practical consequences are worth naming plainly. A multinational estate very easily ends up with two or more signage conventions in one building, typically because equipment was imported with its manufacturer's labelling while the building signage followed the local system. A sign bought from a supplier in another market may be perfectly compliant there and non-compliant where it is installed. Symbol-only signage and signal-word signage cannot simply be swapped for one another, because one assumes the viewer needs no language and the other assumes the viewer reads the wording.

5. Symbols versus text, honestly

The argument for symbols is strong and mostly correct. A pictogram crosses language barriers, is recognised faster than a sentence, and survives distance and poor light better. On a site with a workforce drawn from a dozen first languages, which describes most construction and facilities operations in the Gulf, symbol-led signage is clearly the right instinct.

The honest qualification is that symbol comprehension is regularly worse than the people specifying the signs assume. The handful of very familiar pictograms are understood almost universally. The less common ones are not, and a symbol that is obvious to a safety officer who has seen it a thousand times can be genuinely ambiguous to someone meeting it for the first time. Abstract hazards fare worst; a symbol for a physical object is easier to read than a symbol for a concept. The conclusion is not to abandon symbols. It is that comprehension has to be established rather than presumed, by asking people on site what a sign means and listening to the answer.

Which leads to the point most signage programmes skip: a sign is not a substitute for induction or training. Symbols reduce the language dependency of the reminder, they do not remove the need for the initial briefing that teaches the convention and the specific hazards of the site. Signage is the reminder at the point of use, and it is very good at that. Teaching is done by induction, by the toolbox talk, and by the wider safety management system described in the HSE overview. A useful exercise for a toolbox talk, incidentally, is to walk the team past five signs in their own work area and ask what each one means. The results are usually instructive, and occasionally alarming.

6. Where signage genuinely earns its place

Being sceptical about signage as a control is not the same as being against signage. There are jobs where it is the correct answer and no other control does the work.

• Residual hazards that cannot be eliminated. A hot pipe that must be hot, a step that cannot be removed, a confined space that must exist. Once elimination and engineering have been exhausted, informing people is what is left, and it is legitimate.
• Escape and emergency information. Escape routes, final exits, assembly points, first aid, eyewash, alarm call points. As set out above, this is signage doing its own job rather than covering for a missing control.
• Identification of equipment and isolation points. This is the most under-rated category in the whole subject and the one closest to my own work. An unlabelled isolation point is a safety problem, full stop. If a technician cannot identify with certainty which valve, breaker or isolator serves the equipment in front of them, every isolation is a guess. Labelling is one of the highest-value and cheapest things a maintenance organisation can do, and it is the practical link between the asset register and the physical plant: the label should carry the same identifier the asset hierarchy uses, so that the sign, the drawing, the work order and the isolation certificate all name the same thing. See the lockout tagout guide for how isolation depends on that identification.
• Tags on isolated equipment. Tags are signage, and they are the one category of sign whose whole purpose is to be temporary and to be attached to a specific act by a specific person. The content, control and limitations of tags are their own subject, treated in the LOTO tags guide rather than here. The connection worth holding onto is that a tag warns and does not restrain, which is the same weakness every sign has.
• Pipework and services identification. Knowing what is in a line before breaking into it is a genuine hazard control, and it depends on labelling schemes that survive decades and refurbishments.
• Restricted areas. Where access is controlled, signage at the boundary tells people the boundary exists and what the condition of entry is. It works best alongside a physical barrier, not instead of one.
• Temporary hazards from work in progress. The most legitimate use of signage and the one handled worst, which is why it gets its own section below.

7. When a sign is the right control and when it is standing in for one

This distinction is worth making explicit, because it is the question a safety review should be asking of every sign on the wall.

A permanent sign warning of a hazard that could be fixed is worth reading as a piece of evidence. It tells you the organisation knew about the hazard, chose the cheapest response, and left it there. Any competent risk assessment should catch that, and the hazard identification process should be treating a long-standing warning sign as a prompt to revisit the control rather than as a closed action.

8. Temporary signage, where signage fails most visibly

Temporary signage is the category that does most damage to a signage programme, and it does the damage in three specific ways.

Signs left up after the work finished. A barrier and a warning sign go up for a two-day job, the job runs over, the sign stays, and three months later it is part of the scenery. This is worse than having put up no sign at all, because it teaches every person who walks past that signs on this site do not correspond to reality. Once that lesson is learned, the next genuine warning is discounted too. There is no technical fix; the fix is that whoever puts a temporary sign up owns its removal, and permit close-out includes restoring the signage to its permanent state. That is a natural hook in a permit system, and it is one of the small checks worth building into whatever workflow manages permits, as discussed in the permit-to-work integration guide.

Signs that contradict the permanent signage. A temporary notice diverting escape traffic one way while the permanent escape signage still points the other way is a serious condition, not an untidy one. Any temporary signage that touches escape routes needs deliberate management of the permanent signage it overrides, including covering or amending it and restoring it afterwards.

Signs and tape used where a barrier was needed. Barrier tape communicates; it does not restrain anybody. A floor opening, an excavation or a live-edge working area needs something physical. Tape and a sign around an open riser is the clearest example in the whole subject of an administrative control standing in for an engineering one, and it usually happens not out of ignorance but because the tape was in the van and the barrier was not.

9. Sign blindness and clutter, the argument nobody wants to hear

This is the most useful thing on this page, and it runs directly against the instinct most organisations have: the effectiveness of signage is inversely related to its quantity.

A wall carrying fifteen notices is read by nobody. Adding a sixteenth does not add a sixteenth message; it slightly reduces the readability of the other fifteen. Attention is finite, and a dense field of similar-looking notices is processed as texture rather than as information. Anyone who has watched people walk through a plant-room door plastered with signs has seen this: the eyes do not stop. The signs are present, they are compliant, and they are invisible.

The mechanism by which this happens is entirely understandable. After an incident, adding a sign is cheap, immediate and demonstrable. It closes the corrective action, it appears in the report, and it requires no capital, no design change and no negotiation with operations. Do that after every incident for ten years and the accumulated result is a site where signage carries less information than it did at the start, achieved one reasonable decision at a time.

10. When a sign is the wrong answer

Three patterns are worth recognising by name, because they recur in incident reports everywhere.

The sign as a response to something that needed engineering. An injury occurs at a pinch point, the corrective action is a warning label, and the pinch point remains. The hierarchy exists precisely to stop this substitution, and referring back to the hierarchy of controls during corrective-action review is the practical defence. A sign may be a legitimate interim measure while a guard is designed and procured, but it should be recorded as interim, with the engineering action still open.

The sign as a transfer of responsibility. This is the uncomfortable one. A sign creates a record that the organisation said something, which makes it tempting to use signage as a way of moving responsibility onto whoever walks past it: we told them, it was on the wall, they proceeded anyway. That reasoning is how an organisation ends up investigating the behaviour instead of the condition. The unsafe act versus unsafe condition distinction is the right corrective lens here: if the condition was foreseeable and controllable, a sign warning about it does not convert it into somebody else's unsafe act.

The sign as a substitute for briefing. A complex or task-specific instruction does not fit on a sign and will not be absorbed from one. If the message needs a paragraph, it needs a person. Signs are for short, immediate, at-the-point-of-use messages, and asking them to carry procedure is asking them to fail.

11. Maintaining signage, which almost nobody writes about

Signage is an asset class with a condition, and treating it as a one-off installation is why so much of it is wrong. Signs fade in sunlight until the graphic is barely distinguishable. They are painted over during redecoration, which happens far more often than anyone admits. They are obscured by racking, pallets, a new distribution board or a contractor's stack of materials, and an obscured escape sign is a non-functional escape sign. They become factually wrong when the plant changes: equipment removed, a room repurposed, a route rerouted, an isolation point relocated. And photoluminescent escape signage degrades with age, so escape signage that looked adequate under normal lighting can be inadequate in the dark condition it exists for. Whether any of that meets a requirement is a question for the adopted standard and the local authority, not for this article; that it needs inspecting is a maintenance fact.

So a signage survey is a real task, and the triggers are plain: after any alteration to layout, plant or occupancy; after any change to escape routes; periodically alongside other building-fabric condition rounds; and at handover on any project. The survey I would advise is a walk with a plan, recording each sign's location, category, condition, legibility from the actual viewing position, and whether it is still true. That last column finds most of the problems. Logging signage as inspectable items sits naturally within whatever system already carries the PPM rounds, and the facilities maintenance management guide covers how those rounds are structured.

12. How signage programmes fail, and the response

Two of these deserve a further word. Position is the most common technical error and the easiest to fix: a sign placed after the point at which the viewer had to decide is decoration. If the instruction is to don hearing protection before entering, the sign belongs where the protection is available and before the door, not inside the room. And the hazard-without-action pattern is worth hunting for deliberately, because it feels informative while leaving the viewer with nothing to do.

13. The legal position, in words rather than citations

Most jurisdictions place a duty on employers and on those in control of premises to provide information about residual risks, to mark escape routes, and to identify certain hazards and equipment, and many also require that signage follow a particular national or regional system. What that duty says, which standard it references, and who enforces it differ completely from country to country, and getting it wrong in either direction is a real problem: assuming a foreign rule applies, or assuming none does.

A few things can be said cleanly. The standards named earlier, ISO 7010:2019 as amended and the ISO 3864 parts (both ISO, international, voluntary), and the ANSI Z535 family (United States, voluntary consensus), are not law by themselves anywhere. They acquire force only where a jurisdiction adopts them, where a contract requires them, or where a regulator names them. ISO 45001:2018 as amended by Amd 1:2024 is likewise voluntary and international; it is certifiable, and its clause 8.1.2 requirement makes signage decisions auditable within a certified management system, but certification is not legal compliance. NIOSH, in the United States, has no regulatory power at all. And US or UK requirements have no legal force in the United Arab Emirates, where the binding instruments are the federal employment law and the emirate-level safety frameworks. What your signage must look like comes from the standard adopted in your jurisdiction, in its current edition, with competent advice.

The idea to walk away with

A safety sign is an administrative control that removes no hazard and depends on four fragile conditions being met at once. That makes signage a genuinely useful tool for a narrow set of jobs, informing people of residual risk, telling strangers where the exit is, identifying equipment and isolation points, and a poor substitute for anything higher in the hierarchy. The structure of the subject is the message categories and the shape-and-colour grammar that lets a viewer read the category before the detail. The specific colours, shapes and sizes belong to the standard your jurisdiction adopted, they differ between the major systems, and the worst outcome is not an unread sign but a confidently misread one.

And the single most actionable idea: fewer signs, kept current, beats more signs. If a signage review produces a longer list of signs than it started with, it was an inventory, not a review.

Final thoughts

The two pieces of signage work I would put first in any facilities organisation are the least glamorous. Label the isolation points and the service lines properly, keyed to the same identifiers the asset register uses, because an unlabelled isolator makes every isolation a gamble and the fix costs almost nothing. Then survey what is already on the walls and take down everything that is stale, obscured, faded or no longer true. Both are cheap, both are within a maintenance team's own control, and both do more for the readability of the site's safety information than any new sign will.

After that, get the convention decided and stated, brief people on it rather than assuming the pictograms explain themselves, and put signage on a condition round. Signage will never be a strong control and should never be asked to be one. Treated as a communication system with a maintained condition and a strict attention budget, it does its own job well.

Related reading: Hierarchy of controls, PPE complete guide, LOTO tags, Lockout tagout, Unsafe act vs unsafe condition, Hazard identification, Risk assessment, Toolbox talks, What is HSE, Fire alarm systems, Asset hierarchy design, Facilities maintenance management, Permit to work integration, Life-safety system PM.

Primary sources: ISO , US OSHA , UK HSE , NIOSH .