The bit almost every template site gets wrong
There are two sentences in COSHH 2002 regulation 6 and they do different jobs.
Regulation 6(1) is an absolute prohibition. An employer "shall not carry out work" liable to expose employees to a substance hazardous to health unless a suitable and sufficient assessment has been made and the steps implemented. No headcount threshold. It is a condition of doing the work at all.
Regulation 6(4) is where the five-employee threshold actually lives, and it applies only to recording: "Where the employer employs 5 or more employees, he shall record the significant findings".
So a two-man firm is not legally obliged to file the paperwork, and is absolutely obliged to have done the assessment. Being told the wrong thing about your own duties is how people stop believing the right ones, so this tool says which is which and puts the threshold in the document.
One practical point that needs no false legal claim behind it: writing it down is the only way to show you did it. That is the reason to use this below five employees, and it is enough of a reason on its own.
The check this tool exists for
Regulation 7 sets an order. Exposure must be prevented, or where that is not reasonably practicable, adequately controlled by measures other than PPE, and PPE is used only in addition, where adequate control is not achieved otherwise.
An assessment whose only control is "wear a mask" is therefore upside down. It is also, by a wide margin, the commonest shape of a filled-in COSHH form: hazard identified, mask specified, signed, filed.
So the controls here are grouped by level, in regulation 7's order, numbered, with PPE visibly last and labelled as last. Tick only in the bottom group and you can see that you have only ticked in the bottom group, before anything says so. And the tool refuses to produce the document, rather than warning, because that particular form is not a weak assessment: it is the wrong instrument.
A mask is not a control strategy
Not because masks do not work, but because they only work on a face they have been fit tested on, worn correctly, for the whole exposure, with the right filter, replaced on time. Every one of those is a way for it to fail silently while everybody assumes it is fine.
Water suppression on a cut-off saw does not depend on anybody's discipline. That is the difference the hierarchy is describing.
No exposure limits, and that is permanent
Workplace exposure limits are in EH40/2005, which is revised. A stale limit inside a generated document is worse than no limit at all, because a number on a PDF gets trusted and nobody re-checks it.
We cannot commit to tracking EH40 revisions, so the tool does not carry the figures. It asks whether the substance has a limit, records the answer, and tells you to look it up. That is less impressive and more honest, and it is the same rule we apply to anything we cannot maintain.
Hazard statements work the same way. The H-codes come off your supplier's safety data sheet, not from a product name, because the same trade name covers different formulations and the supplier's statement is not ours to guess.
Mild steel welding fume, since 2019
Worth its own section because a lot of fabricators still have the old position in their heads, and this is the one that changed.
IARC classified all welding fume as a Group 1 carcinogen: causing lung cancer, and possibly kidney cancer, with mild steel included. In February 2019 HSE strengthened its enforcement expectation to match, and the key line is that general ventilation does not achieve the necessary control.
What HSE expects: engineering controls such as LEV for all welding indoors, and appropriate RPE for welding outdoors, with RPE also supplementing LEV wherever LEV alone does not adequately control the residual fume.
Which means the open shutter door and the fan in the corner is not an answer, and "it is only mild steel" stopped being a position in 2019.
COSHH assesses work, not products
This is why the first field is the task. The same tin of solvent-based adhesive brushed in an open stairwell and rolled in an unventilated plant room are two different assessments, and the substance has not changed at all.
A folder of assessments filed by product name is assessing the wrong thing. What matters is the substance, the place, the rate, the duration, and who else is standing nearby, which is why the tool asks about the labourer clearing up behind you and not only the person holding the tool.
The starting points, and what they are not
Ten common trade exposures are offered as a starting point: silica from cutting, welding fume, wood dust, cement, solvents, isocyanates, lead, bitumen, used oils and cleaning chemicals. Picking one seeds the routes of exposure and a sensible set of controls, and shows you what makes that exposure different from a generic dust.
It is not a substance database and it is not offered as one. What is well established regardless of brand is the control approach for a familiar job, and none of the ten seeds PPE alone, which is enforced by a test rather than left to good intentions.
Face fit testing, and the beard
Tick a disposable or reusable mask and the tool asks whether it has been fit tested, because RPE that seals to the face gives protection nobody can quantify until it has, and gives none at all over stubble.
That is not a preference about facial hair. A tight-fitting mask works by sealing, and it cannot seal against hair. Somebody who needs a beard needs a powered hood instead, which is a different purchase and a conversation worth having before the job rather than during it.
The substance here, the task next door
A COSHH assessment covers the substance. The risk assessment covers the rest of the task, and RAMS puts it with the method statement if the principal contractor wants both. Silica, dust and fume also belong in the construction phase plan, which HSE lists them in by name.
Nothing is uploaded and nothing survives closing the tab, so download the PDF before you shut it.
Common questions
Do I have to write a COSHH assessment down?
Only if you employ five or more people. Regulation 6(4) of COSHH 2002 requires the significant findings to be recorded "where the employer employs 5 or more employees", and not below that. Most template sites tell everybody they must record it, which is wrong for a two-man firm and is how people stop believing the parts that are true.
So a small firm does not need a COSHH assessment at all?
No: that is the other half, and it is the more important half. Regulation 6(1) says an employer "shall not carry out work" liable to expose employees to a hazardous substance unless a suitable and sufficient assessment has been made and the steps implemented. No headcount. It is a condition of doing the work at all. The assessment is always required; only the paperwork has a threshold. And writing it down is still the only way to show you did it.
Why does it refuse when I only tick masks and gloves?
Because that is an upside-down assessment. Regulation 7 requires exposure to be prevented, or where that is not reasonably practicable, adequately controlled by measures other than PPE, with PPE used only in addition, where adequate control is not achieved otherwise. An assessment whose only control is "wear a mask" has the hierarchy inverted, and it is the single commonest shape of a filled-in COSHH form.
Why are there no exposure limits in it?
Deliberately, and permanently. Workplace exposure limits live in EH40/2005, which gets revised, and a stale limit inside a generated document is worse than no limit because it gets trusted. The tool asks whether the substance has one and points you at EH40 to look it up. Same reason we do not ship tax rates we cannot commit to maintaining.
Is mild steel welding fume really a problem now?
Yes, and this one caught a lot of people out. IARC classified all welding fume as a Group 1 carcinogen, mild steel included, and in February 2019 HSE strengthened its enforcement expectation accordingly. General ventilation does not achieve the necessary control, so HSE expects engineering controls such as LEV for all welding indoors, with appropriate RPE for welding outdoors. There is no "it is only mild steel" any more.
Why does it ask for the task rather than just the product?
Because COSHH assesses work, not products. The same tin of solvent-based adhesive is a different assessment brushed in an open stairwell and rolled in an unventilated plant room, and the substance has not changed at all. An assessment filed against a product name rather than a job is assessing the wrong thing.