A lone worker is anyone who does their job out of sight and earshot of a colleague, the off-shift maintenance tech, the night-crew operator, the worker in a remote corner of the plant, so that if they are hurt or overcome, no one nearby knows to help. A lone worker safety program manages that isolation with a risk assessment, reliable check-ins, and limits on which tasks may be done alone at all.
The danger with lone work is not that solo tasks are inherently more hazardous. It is that the normal backstop, someone noticing you are in trouble, is gone. A twisted ankle is a nuisance with a coworker present and a genuine emergency at 2 a.m. in an empty warehouse. This guide covers how to assess lone-work risk, build check-ins that actually work, use man-down and duress technology, and draw the line on tasks that should never be solo. It is educational, not legal advice.
Does OSHA regulate lone workers?
Not with a single dedicated standard for general industry. There is no “lone worker rule” you can point to the way you can point to lockout/tagout. The closest specific standard is 29 CFR 1915.84 which requires shipyard employers to check on employees working alone at regular intervals. For everyone else, the obligation flows from the General Duty Clause: employers must furnish a workplace free of recognized hazards likely to cause death or serious physical harm, and working alone with a serious hazard and no way to summon help is exactly that.
Several task-specific standards also constrain solo work indirectly. Confined space entry under 1910.146 requires an attendant outside the space, a permit-space entrant is never truly alone by rule. Respiratory-protection situations and certain atmospheres like hydrogen sulfide effectively demand a buddy. So while there is no blanket lone-worker mandate, the practical picture is that OSHA expects you to have looked at the isolation and done something about it.
How do you assess lone-worker risk?
Start by naming who actually works alone and when. It is usually more people than a manager first guesses: the maintenance tech who stays late, the shipping clerk who opens at 4 a.m., the quality lab running a solo weekend shift, the driver between stops. Then ask, for each, one blunt question: if this person collapsed or was injured right now, how long until anyone noticed, and how bad would it get in that time?
That question sorts your lone workers into risk tiers and sets the check-in interval. A worker doing light office tasks with a phone in their pocket is low risk. A worker near moving equipment, chemicals, heat, or height is not. The assessment feeds straight into the job safety analysis for the specific solo tasks, and the same discipline applies to contractors who work your site alone off-hours.
What makes a check-in system work?
A check-in system is only as good as the failure it triggers. The mechanics are simple: the lone worker confirms they are fine on a set interval, and if a check-in is missed, a defined escalation kicks off, a call, then a second contact, then someone physically dispatched to look. The interval is set by the risk assessment, not by convenience. A worker near serious hazards might check in every 15 or 30 minutes; a low-risk worker, hourly.
Three details separate a real system from a comforting one. It needs a backup contact, so a missed check-in does not die in one unanswered phone. It benefits from a duress signal, a word or a button that means “send help now” without alerting a threat. And the escalation must have a name attached and a maximum time: “if no response in 10 minutes, the shift lead drives to the location” beats “someone will check.” A check-in schedule with no one accountable for the missed call is theater.
When should you use man-down and duress technology?
Technology earns its place where check-in intervals leave dangerous gaps. Man-down devices detect a fall, an impact, or a period of no motion and raise an alarm automatically, useful precisely because an injured worker may be unable to call in. Duress buttons let a worker signal trouble instantly. Location beacons tell responders where to go, which in a large plant is half the rescue. Two-way radios and satellite messengers cover areas with no cell coverage.
Choose the technology to the gap, not the catalog. A man-down alarm on a worker who is never more than a few minutes overdue adds little; the same device on a solo tech in a boiler room at night can be the difference between a bruise and a fatality. And technology never replaces the rules below, a duress button does not make it acceptable to enter a permit space alone.
How do you build a lone worker program?
Assemble it in order, so each piece rests on the one before it. This is the single framework to follow.
- Identify every lone worker and when they are alone. List roles, shifts, and locations. Include the people who are alone only occasionally, they are the ones nobody plans for.
- Assess the risk of each solo situation. For each, judge how fast a problem turns serious and how long until anyone would notice. That sets the tier.
- Set the controls for each tier. Assign a check-in interval, decide whether man-down or duress technology is warranted, and define the escalation with names and time limits.
- Ban the tasks that should not be solo. Confined space, energized electrical or lockout/tagout servicing on hazardous energy, work at height, and work in atmospheres like hydrogen sulfide are two-person jobs by policy, period.
- Train, drill, and review. Train workers and the responders who receive the alerts, run a drill on a missed check-in, and revisit the plan after any incident, near miss, or change in staffing.
Step five is where most programs quietly fail: they write the escalation and never test it. Run one drill where a check-in is deliberately missed and time how long until someone actually reaches the worker. Feed what you learn into near-miss reporting and make sure the program lines up with your emergency response so a real man-down alarm connects to the people who can act.
Which tasks should never be done alone?
Some jobs carry a failure mode so fast or so severe that no check-in interval is short enough. Confined space entry is the clearest: an atmosphere can incapacitate an entrant in seconds, which is why the standard requires an attendant outside. Energized electrical work and arc flash-exposed tasks belong on the list, as does work at significant height where a fall arrest still leaves a worker suspended and needing prompt rescue. So does entry into oxygen-deficient or toxic atmospheres. For these, the control is not a better phone, it is a second qualified person present. Writing that ban into policy is the single highest-value line in a lone-worker program.
The reason a check-in cannot save these tasks is the speed of the harm. A fifteen-minute check-in interval means, on average, help learns of a problem seven or eight minutes after it starts and arrives minutes after that. For a sprain in a warehouse, that delay is survivable. For an entrant overcome by a low-oxygen atmosphere, or a worker in contact with an energized conductor, minutes are the whole event, the outcome is decided before the first missed check-in is even noticed. That is why the answer for these jobs is presence, not monitoring: someone standing by who can cut power, pull an entrant on a retrieval line, or start rescue in the seconds that matter. Draw the line in writing, name the specific tasks, and make it a rule a supervisor cannot waive to keep a job moving.
What do the numbers say?
The regulatory anchors and the stakes, from primary sources:
- The shipyard “working alone” requirement to check on lone employees at regular intervals is at 29 CFR 1915.84.
- The attendant requirement that keeps permit-space entrants from working truly alone is in the confined spaces standard, 29 CFR 1910.146.
- The Bureau of Labor Statistics counted 5,283 fatal work injuries in 2023 and 5,070 in 2024; isolation turns a survivable injury into a fatal one when no one is there to respond.
The math of lone work is simple and unforgiving: the same injury is more likely to kill someone who is alone, because rescue is delayed. A program does not make solo tasks safe, it shortens the time between “something went wrong” and “someone is helping.”
That time depends on information reaching the right person fast, which is hard when check-ins live on a clipboard and escalation lives in one supervisor's head. Harmony is an AI-native layer that connects machines, software, and paperwork into one operational layer, with no rip-and-replace, so check-in logs, escalation contacts, and the record of who is working where become structured and searchable rather than scattered, the everyday shape of connected worker technology. Harmony's digital workflows move handoffs and status into that structure; it is not a lone-worker monitoring product, but it keeps the information where the response is. See the CLS case study.