What’s a Fire Drill? Meaning and How To Conduct It?

A fire drill is a rehearsed evacuation of a building, started by the alarm and run as though the fire were real, to test whether the evacuation strategy actually works with the people and conditions you have on site that day. That is the definition. The part most organisations get wrong is the purpose: a drill is not a training exercise you pass, it is a measurement you take.

In fifteen years across fire protection work in fourteen countries — including two years as Emergency Drill Realism Lead at NAFFCO International and eighteen months running muster assurance for Kidde Fire Australia Partners — the number I see recorded on drill sheets is almost always the same one: total evacuation time. It is the least diagnostic figure on the page. This article covers what a fire drill is, what the law in the US and UK actually requires (which is not what most guidance claims), how to run one, and which numbers tell you something you can fix.

Key takeaways

Six points govern everything below, and two of them contradict the guidance ranking above this page:

• OSHA does not require fire drills. 29 CFR 1910.38 sets six minimum elements for an emergency action plan; drill frequency is not among them. If your plan says you will drill, OSHA enforces that commitment.
• The UK Fire Safety Order sets no drill frequency either. Article 15(1)(a) requires appropriate procedures "including safety drills" — the annual figure comes from guidance and BS 9999, not from law.
• US frequency comes from your fire code, by occupancy. Under IFC Table 405.2, offices below 500 occupants often have no mandated drill at all, while schools drill monthly and hospitals quarterly on each shift.
• Total evacuation time fuses two different problems. Pre-movement time — the gap between alarm and first movement — is the part a drill can actually change. Travel time is a building constraint.
• A drill that everyone expects measures compliance, not capability. NFPA 101 §4.7.3 asks for drills at unexpected times and under varying conditions.
• Accountability is where drills fail quietly. A roll call against a stale roster returns "all clear" whether or not it is true.

What a fire drill is — and what it is actually testing

A fire drill puts the alarm signal into a normally occupied building and requires occupants to leave by their designated routes to a place of safety, where they are accounted for. Everything else — the timing, the observers, the debrief — exists to turn that event into evidence.

What the drill tests is not people's willingness to walk outside. Almost everyone will walk outside. It tests four things they cannot self-report: whether the alarm is audible and intelligible everywhere someone might be, whether the routes work when loaded, whether the wardens sweep the areas they are assigned, and whether your accountability system can tell you the difference between "everyone is out" and "everyone we know about is out."

That last distinction is the one worth holding onto. During a warehouse fire controls assignment with Siemens Building Technologies, the recurring finding was not blocked exits — it was that mezzanine and plant-room areas sat outside every warden's assigned sweep because the sweep map had been drawn from the office floor plan.

Fire drill vs alarm test vs evacuation exercise

These three get conflated in policies and then in records, which causes real problems when an assessor asks how many drills you ran last year.

An alarm test is not a drill and cannot be logged as one. Equally, an unplanned evacuation caused by a genuine alarm activation is not a substitute for a required drill — several US fire codes state this explicitly.

The distinctions that matter when you write your programme are:

• A drill requires occupant movement. If nobody left the building, it was a test.
• A drill requires accountability. Walking out without a roll call gives you a fire alarm, not a drill.
• A drill requires a record. An unrecorded drill did not happen as far as any assessor is concerned.

What the law requires — and the two things it does not say

Two claims appear on nearly every page ranking for this topic, and both are wrong. The first is that OSHA requires an annual fire drill. The second is that the UK Fire Safety Order requires one. Neither instrument says anything of the kind, and if you have built your programme on either belief, the frequency you are working to came from somewhere else entirely.

This matters beyond pedantry. If you think the law hands you a number, you stop asking what number your building needs.

The three questions to settle before you commit to a frequency are:

• Which instrument actually binds this building? In the US that is the adopted fire code, not OSHA.
• What does your own emergency action plan already promise? That promise is enforceable even where the standard is silent.
• Can your fire risk assessment justify the interval you chose? Both jurisdictions put the burden there.

United States — 29 CFR 1910.38 and the missing frequency

An employer must have an emergency action plan whenever another OSHA standard in Part 1910 requires one — and where a written plan is required, the six minimum elements at 1910.38(c) cover reporting procedures, evacuation procedures and exit route assignments, procedures for employees who stay behind to run critical operations, procedures to account for all employees after evacuation, rescue and medical duties, and a named contact for questions about the plan.

Drills are not in that list. The standard also requires an employee alarm system under 1910.38(d) and requires the employer to designate and train employees to assist in a safe and orderly evacuation under 1910.38(e), but it sets no drill interval.

What OSHA will enforce is your own plan. If the EAP commits you to quarterly drills, an inspector can cite the failure to run them. This is the practical trap: an ambitious plan written once and never resourced creates an obligation that did not previously exist.

Do not read the absence of a rule as permission to skip drills. OSHA's stated position is that practice drills are a good idea as often as necessary to keep employees prepared, and where an incident occurs that drills would have prevented, the agency can fall back on the General Duty Clause. "No specific standard" is not the same as "no exposure."

United Kingdom — Article 15 of the Fire Safety Order

The Regulatory Reform (Fire Safety) Order 2005 requires the responsible person to establish and, where necessary, give effect to appropriate procedures, including safety drills, to be followed in the event of serious and imminent danger, and to nominate a sufficient number of competent persons to implement those procedures so far as they relate to evacuation.

Article 15(3) then defines competence for this purpose: sufficient training and experience, or knowledge and other qualities, to properly implement the evacuation procedures. That is a functional test, not a certificate — your wardens qualify by being able to do the job, and you should be able to show why you think they can.

Read the duty wording carefully. "Where necessary" and "appropriate" put the frequency question back on your fire risk assessment. The Order does not say annually. The Home Office guidance for responsible persons and BS 9999 are where the annual convention comes from, and guidance is what an inspector will measure you against — but the legal duty is to justify your interval, not to hit a fixed one.

Where the frequency numbers actually come from

In the US, the binding numbers sit in the adopted fire code, not in OSHA. Under IFC Table 405.2 and the parallel NFPA 101 occupancy chapters, the interval depends entirely on how your building is classified.

Source: International Fire Code Chapter 4, Section 405 and Table 405.2, as adopted and amended locally. Verify against your jurisdiction's adopted edition — states amend these tables routinely.

The Group B footnote is the one worth reading twice. A 200-person office in a low-rise building has, in many US jurisdictions, no code-mandated fire drill whatsoever. Whether you drill it is a risk decision, and I would still drill it — but you should know you are choosing to, not complying.

⚖️ Jurisdiction note: Frequency is set locally in almost every country. Australia works to AS 3745, Canada to provincial fire codes under the NBC/NFC framework, and the Gulf states typically to a civil defence code layered over NFPA. Route the requirement through the adopted code where the building sits, not the head office's home code.

Before the drill: the decisions that determine what you learn

Most of a drill's value is decided before the alarm sounds. A drill designed to go smoothly will go smoothly and teach you nothing. A drill designed to surface a specific weakness will surface it, and you will spend the next fortnight fixing something real.

Write down, before the date, the question this drill is answering. "Can we evacuate?" is not a question — you already know the answer. "Can the night shift evacuate the east warehouse when the main roller door is the blocked route?" is.

Scenario design — blocking a route, and why

NFPA 101 §4.7.3 asks that drills be held at expected and unexpected times and under varying conditions to simulate the unusual conditions that occur in an actual emergency. Fire does not politely leave your primary exit available.

The practical method is to nominate one exit as unavailable and place an observer there to redirect people, rather than physically obstructing it — never block a real exit for a drill. What you are testing is whether occupants have a second route in their heads, and whether the secondary route has the capacity for the load it suddenly receives.

Who to notify, and who not to

The notification list is where drills quietly become announced when you meant them to be unannounced. Wardens and observers need to know. The alarm receiving centre and the fire and rescue service need to know, so nobody is dispatched. Building management and co-tenants need to know in a shared building.

General occupants do not. If your policy is to warn everyone a week ahead, you have chosen to measure the best case, and pre-movement time in that drill will bear no relationship to a real one.

Observer placement

Observers standing at the assembly point see the end of the story. Position them where the failures happen instead.

Before the drill date, assign observers to these positions and brief each on what to record:

• One per stair core or main route, recording flow, hesitation and counterflow
• One in the most acoustically isolated area you have — plant rooms, cold stores, toilets, loading docks
• One shadowing a warden through a full sweep, timing it
• One at the muster point with a stopwatch and the roster
• One at the emergency vehicle access route, checking it is clear

How to conduct a fire drill, step by step

The sequence below is the one I use for assurance walkdowns, ordered by the drill's own clock rather than by the planning cycle. Each step has a thing to do and a thing to watch.

1. Activate the alarm — and watch the first 30 seconds

Start the drill by activating the fire alarm system from a manual call point, not by announcement. Several fire codes require this where a system exists, and it is the only way to test detection-to-notification as occupants experience it.

The first 30 seconds is the most informative part of the whole exercise. Watch for people finishing sentences, saving files, looking at colleagues to see what they do, and asking whether it is real. That behaviour is the pre-movement phase, and it is where your minutes are going.

2. Sweep and warden discipline

Wardens check their assigned areas and move to the muster point, reporting their area clear. Watch whether they actually enter the isolated spaces or glance down the corridor, and whether anyone sweeps an area that is not on anybody's list.

3. Muster and accountability

Occupants gather at the designated assembly point — the specific location in the plan, well clear of the building and of the fire service approach. Roll call runs against a current roster that includes visitors and contractors.

4. All-clear and re-entry

One named person authorises re-entry, after wardens have reported and accountability is resolved. The recall signal must be distinct from the evacuation signal; IFC 405.9 requires exactly this where an electrical or mechanical recall signal is used, and reusing the same tone teaches occupants to ignore it.

The four things I check on a drill walkdown, in this order, are:

1. Was the alarm audible and intelligible in the quietest occupied space? Ask someone who was there, not the panel.
2. Did the sweep cover every space on the floor plan, or every space on the warden's habit?
3. Was the roster current as of that morning? Not last month.
4. Did one named person authorise re-entry? If people drifted back in, you have no all-clear discipline.

The numbers worth recording — and the one everyone records instead

Fire engineering splits the time an evacuation needs into components. Required safe egress time (RSET) is the sum of detection time, alarm time, pre-movement time, and travel time, and it is compared against available safe egress time (ASET) — how long before conditions become untenable. The framework sits in PD 7974-6, the human factors part of the BS 7974 fire engineering series, and underpins the evacuation assumptions in BS 9999.

Pre-movement time is the phase covering recognition, decision-making and preparation, and in most occupancies it is longer and far more variable than the walking itself. Which leads to the point of this section: a single stopwatch figure for "evacuation time" adds together the one thing you can change and the one thing you cannot.

Travel time is a function of distance, exit width and occupant numbers. Short of construction, you are stuck with it. Pre-movement time is a function of alarm intelligibility, occupant familiarity, warden assertiveness and how seriously the organisation is known to take alarms — all of which respond to management within weeks.

If the drill clears in under 90 seconds, that is usually not a good result — it generally means people took a familiar shortcut rather than the designated route, or did not walk the full distance to the muster point.

The four figures worth putting on the drill sheet are:

• Pre-movement time per area, not per building — the isolated areas are the outliers you need
• Sweep completion time per warden, which tells you whether the areas are sized realistically
• Accountability time, from last arrival to confirmed roll call
• Number of people who used a route other than their designated one, which is the cheapest single indicator of whether the plan matches reality

What goes wrong — findings from the muster point

The failures below are the ones I write up most often. None of them appear in the evacuation time. All of them appear in the debrief if somebody is standing in the right place.

Emergency access blocked by the site's own work

On a fire life safety programme assignment for Wormald Fire Systems in Brisbane, I found equipment staged across the emergency vehicle access route during a live walkdown. Not abandoned — staged, deliberately, by people doing their jobs, in the only clear space available to them that morning.

I stopped the pattern, had the route reinstated, and briefed the crew before work restarted. Follow-up observations on that sector showed fewer repeats.

The transferable point is that the finding is a logistics problem wearing a fire safety label. Nobody decided to block the fire lane. Somebody decided where to put a delivery, and the fire lane was the space that was free. If your drill does not put an observer on the approach route, you will not see it — and the fire service will discover it for you, once.

The off-shift gap

Across several assurance programmes I have run night-shift control integrity work, and the pattern repeats: evacuation arrangements are validated against the day shift and inherited by everyone else. At night there are fewer wardens, more lone working, more contractors, and a building the occupants navigate mostly by memory.

If every drill on your log ran on a Tuesday morning, your evacuation strategy is unvalidated for most of the hours your building is occupied.

Accountability that counts the wrong people

A roll call against a roster printed on Monday returns "all clear" on Thursday regardless of the truth. The failure modes are consistent: visitors signed in at reception but never added to the muster list, contractors inducted at a separate gate, staff who left the site at lunch and are counted missing, and staff working from another floor counted twice.

The three checks that catch most accountability failures are:

• Reconcile the roster against the access control log for that morning, not against the HR list
• Give contractors a muster point and a named person, and check they know both
• Time the roll call separately — accountability taking eight minutes is a finding even if the walk out took two

Evacuating people who cannot use the stairs

This is the part of the drill most often simulated rather than performed, and simulation is where the plan stops being tested. A personal emergency evacuation plan (PEEP) that has never been walked is a document, not an arrangement.

The reasonable middle ground: do not require someone to be carried down six flights on a quarterly basis. Do require, at least annually, that the assigned assisters locate the equipment, deploy it, and move the person the first flight — or that the refuge area's two-way communication is used and answered.

Refuge areas are where I find the most consistent gaps. During egress integrity work with Chubb Fire & Security AU, the recurring finding was refuge points whose call systems had never been answered by a named role — the equipment worked, and there was nobody with a documented duty to respond to it.

The checks worth building into every drill involving assisted evacuation are:

• Evacuation chairs deployed, not just pointed at — including by the specific people rostered that day
• Refuge call answered by a named role, with the response time recorded
• The assister's own evacuation accounted for — an assister who stays is now a person needing rescue
• Temporary needs captured — a broken ankle, late pregnancy or a migraine changes the picture that week

The drill record — what an assessor reads

A drill record has two audiences: the person who will run the next drill, and whoever investigates after a real fire. Write it for the second one.

The most complete published specification is IFC 405.5, which lists what a drill record must contain. It is worth adopting whatever jurisdiction you sit in, because it is specific enough to be useful rather than a box to tick.

Source: International Fire Code Section 405.5. Verify against your locally adopted edition.

I would add two fields the code does not require: pre-movement time per area, and the corrective action owner with a due date. A problem logged without an owner is a problem logged twice.

The record failures that cost organisations most after a real fire are:

• Drills logged with no problems encountered — a flawless record across years reads as a record nobody wrote honestly
• Corrective actions with no owner, which recur in the next drill and the one after
• No shift or scenario variation visible in the dates and conditions columns
• Times recorded to the minute when the useful resolution is seconds

How often to run fire drills

Start from the adopted fire code for your occupancy, then ask whether that number is enough for your building. In most cases the code figure is a floor set for a generic version of your occupancy, and your risk assessment has information the code writer did not.

Four conditions push the interval shorter than the code baseline: high staff turnover, occupants who sleep or cannot self-evacuate, a building whose layout has changed, and a workforce that hears alarms often enough to have learned to ignore them. That last one is the most underrated. A site with frequent false alarms has a pre-movement problem no amount of signage will solve.

Rather than a single annual event, spread the programme so it covers your actual operating pattern:

• Rotate the shift — a drill on every shift over a two-year cycle, not five Tuesday mornings
• Rotate the scenario — a different route unavailable each time
• Rotate the season — a February muster is a different exercise from a July one
• Drill after every change — layout, occupancy, warden turnover, or alarm system work
• Re-drill after a poor result — within weeks, not at the next annual slot

Frequently asked questions

How often should a fire drill be carried out?

It depends on your occupancy and adopted fire code. Under IFC Table 405.2, schools drill monthly, hospitals and hotels quarterly on each shift, assembly occupancies quarterly, and qualifying offices annually. UK practice defaults to annually under BS 9999 and Home Office guidance, with more for high-risk or high-turnover premises.

Does OSHA require fire drills?

No. 29 CFR 1910.38 requires an emergency action plan with six minimum elements and sets no drill frequency. However, if your emergency action plan states you will conduct drills, OSHA will require you to conduct them at the cadence your plan commits to.

How long should a fire drill take?

There is no target time — the right figure is the one your building's escape route design assumed. Most single-storey workplaces clear in under three minutes; a five-storey office typically runs four to six. A clearance under 90 seconds usually means people cut the route rather than walking to the muster point.

Should fire drills be announced or unannounced?

Both, over time. NFPA 101 §4.7.3 asks for drills at expected and unexpected times under varying conditions. Announce the first drill after a procedure changes so people learn it correctly; run later ones unannounced so you measure real pre-movement behaviour rather than rehearsed behaviour.

Can an employee refuse to take part in a fire drill?

Participation is part of the safety training an employer is entitled to require, and refusal is handled under your normal training policy. Nobody should be physically compelled, but an employee who will not evacuate in a drill is a live safety problem worth addressing directly rather than disciplining and moving on.

Who is responsible for conducting a fire drill?

In the UK, the responsible person under the Fire Safety Order, who must nominate a sufficient number of competent persons to implement evacuation procedures. In the US, the duty sits with the employer under the emergency action plan and with the building owner or operator under the fire code. In practice one named coordinator runs the drill and wardens execute it.

What should be recorded after a fire drill?

At minimum the nine fields in IFC 405.5: who conducted it, date and time, notification method, staff participating, occupants evacuated, conditions simulated, problems encountered, weather, and evacuation time. Add pre-movement time per area and a named owner with a due date for each corrective action.

Key points

A fire drill is a measurement. Treated that way, it tells you which of your assumptions about evacuation are wrong while it is still cheap to find out.

The legal position in both the US and UK is more open than most guidance suggests: neither OSHA's emergency action plan standard nor the Fire Safety Order sets a frequency, which means the number you drill to is yours to justify. Get the frequency from your adopted fire code, then ask whether it is enough for the building you actually have.

And record the right numbers. Total evacuation time is the figure everyone writes down and nobody can act on. Pre-movement time, sweep completion and accountability time each point at something you can fix before the next drill.

If you take one thing into your next one: put an observer where the failure will be, not at the muster point where the story ends.

The five actions worth taking before your next drill are:

1. Check what your emergency action plan or fire risk assessment already commits you to — that is the interval you are held to.
2. Look up your occupancy group in the locally adopted fire code and confirm the baseline.
3. Write down the question this drill answers before choosing the date.
4. Assign observers to the five positions, not to the muster point.
5. Add pre-movement time and a corrective action owner to your drill record template.

About the author — Mia Collins

Mia Collins is an Australian Fire Life Safety & Emergency Response Assurance Consultant with 15 years of continuous field experience across industrial fire protection, high-rise fire life safety, warehouse fire systems, detection and alarm, muster and accountability, and egress integrity. Her focus is practical assurance — checking that controls still work where work actually happens. She served as Emergency Drill Realism Lead at NAFFCO International and as Muster Assurance Lead at Kidde Fire Australia Partners, and now leads Collins Fire & Emergency Assurance in Brisbane, Australia, after senior roles with Wormald Fire Systems, Chubb Fire & Security AU, Tyco Fire Protection Services AU and Siemens Building Technologies Fire AU.

Credentials: NEBOSH International General Certificate; ISO 45001 Lead Auditor; ISO 14001 Internal Auditor; IOSH Managing Safely; Emergency Response Planning Awareness; Incident Investigation (ICAM pathway).

Sources

• 29 CFR 1910.38 — Emergency action plans, Occupational Safety and Health Administration
• OSHA eTool — Evacuation Plans and Procedures, Occupational Safety and Health Administration
• Regulatory Reform (Fire Safety) Order 2005, Article 15 — legislation.gov.uk (linked in the text above; current to 28 August 2026)
• Does OSHA require fire drills? — J. J. Keller Compliance Network, summarising OSHA's position
• International Fire Code, Chapter 4, Section 405 and Table 405.2 — emergency evacuation drills, frequency and records (verify locally adopted edition)
• NFPA 101 Life Safety Code, Section 4.7 Fire Drills, and occupancy chapters
• ASET and RSET in fire engineering — overview of the PD 7974-6 human factors framework
• BS 9999 and BS 5839 — UK fire safety design, management and alarm system practice