Your Shift Doesn't End at a Time. It Ends When Someone Relieves You.

Look at your schedule. There is a number where your shift ends.

For a lot of jobs, that number is a fact. For a lot of shift work, it is a floor — the earliest the shift could possibly end, assuming everything downstream of you goes right. Whether it actually ends there depends on something that isn't printed anywhere on the schedule: whether someone showed up to take your place, and how long it takes to hand the work over.

This is not the same conversation as being asked to stay late. Mandatory overtime is a decision somebody makes. This is different — it's built into the shape of the job, and it happens on nights when nobody decided anything at all.

The two kinds of ending

Some shifts end when the work is done. You finish, you leave, and if there's more work it's there tomorrow.

Other shifts end when a person arrives. Dispatch consoles, patrol posts, control rooms, staffed apparatus, a nursing assignment, a corrections post — none of these can be left empty. The seat is what's scheduled, not you. So the shift can't end at a time, because a time can't cover a seat. It ends at an event: your relief walking in and taking the work off you.

Once you notice that, a lot of small frustrations stop looking like bad luck:

• The end time varies and the start time doesn't. Nobody is ever relieved early. The distribution only has one tail.
• How late you leave isn't about how hard you worked. It's about how many patients, units, or open items are on the handover — which was decided when the assignment was made, hours before.
• You can't plan around it. Not because it's unpredictable in principle, but because the thing that determines it happens after you've already committed to whatever you scheduled afterwards.

And there's a fourth one that's easy to miss: it can be someone else's clock entirely. A crew waiting to transfer care at a hospital isn't waiting on their own department. Their shift end is being set by a building they don't work for.

"Off duty" has a definition, and it turns on advance notice

Here is where the federal rule is more interesting than most people expect. It doesn't ask whether you were busy. It asks whether you were released.

29 CFR § 785.16(a) says periods when an employee is "completely relieved from duty" and long enough to use for their own purposes are not hours worked — and then adds the condition that does the real work:

He is not completely relieved from duty and cannot use the time effectively for his own purposes unless he is definitely told in advance that he may leave the job and that he will not have to commence work until a definitely specified hour has arrived.

Read that against a relieved post. Being off duty, under this rule, requires being told ahead of time that you may go. "You can leave when your relief gets here" is the opposite structure: you are told you may leave at the moment you may leave. There is no advance, and there is no definitely specified hour. That is the regulation's own description of someone who has not been relieved.

The regulation's example is a truck driver, not a paramedic, but the distinction it draws is the one that matters everywhere: engaged to wait versus waiting to be engaged. Standing on a post you cannot leave, holding work you cannot put down, is the first one.

The handoff itself is work, not a formality

The other half is the handover — report, pass-down, briefing, whatever your service calls it. It happens after the end time, it's often unrecorded, and it is not optional.

The Portal-to-Portal Act interpretation at 29 CFR § 790.8 is the standard here. "Principal activities" includes "all activities which are an integral part of a principal activity," and the regulation extends that to "those closely related activities which are indispensable to its performance." Its own examples are mundane on purpose — oiling a machine, installing a cutting tool, changing into required gear.

A handover is closer to the center of the job than any of those. The next person cannot lawfully or safely take the seat without it. It isn't preparation for the work; it's the mechanism by which the work continues to exist while people change.

Worth noting what the same regulation excludes: "checking in and out and waiting in line to do so" is not integral. That's the actual dividing line — not whether it happens after the end time, but whether it's part of doing the job.

Rounding at the end of a shift is not the same question as rounding at the start

Most timekeeping systems round. That's permitted, and 29 CFR § 785.48(b) explains exactly why:

Presumably, this arrangement averages out so that the employees are fully compensated for all the time they actually work. For enforcement purposes this practice of computing working time will be accepted, provided that it is used in such a manner that it will not result, over a period of time, in failure to compensate the employees properly for all the time they have actually worked.

The permission and its justification are in the same sentence. Rounding is accepted because the errors are assumed to cancel — some days you're rounded up, some days down, and across a year it washes out.

Now apply that to the end of a relieved post. You are never relieved early. The error at that end of the shift doesn't have two directions; it has one. Whatever the rounding interval is, the same seven or nine minutes lands on the same side every single time.

So the two ends of your day are genuinely different questions, even under identical rounding rules. At the start, the assumption behind the rule is at least plausible. At the end of a shift that can only run long, the thing the regulation presumes is happening is, structurally, not happening. That doesn't automatically make a given system unlawful — the test is the outcome over time, not the design — but it does mean the presumption is worth checking rather than assuming.

Which question actually has an answer

"What time do you get off?" is a question built for a job with an end time. On a relieved post, any honest answer needs a conditional attached, and the conditional is the part that sounds like an excuse.

The useful move isn't to find a better answer to that question. It's to swap it for the ones that do have written answers:

1. What releases me? Is the standard "relief is physically present," "handover is complete," or "the end time"? These are three different policies and departments really do differ.
2. Who is responsible when relief doesn't arrive? There is almost always a named holdover or force procedure. It says who calls, in what order, and whether refusing is possible.
3. Is handover time inside the shift or after it? Some schedules build in an overlap — a paid fifteen or thirty minutes where both people are on the clock. If yours does, the end time is real. If it doesn't, the handover has to land somewhere, and it lands on you.
4. How does the timekeeping system round, and in which direction? Not to build a case — just so you know whether the minutes are being captured or absorbed.
5. What's the actual distribution? Note your real leave time for one full rotation. Not to complain: because "usually about twenty minutes" and "usually on time, occasionally ninety" are different problems with different solutions, and most people have never separated them.

The part worth keeping

The reason this stays invisible is that it never looks like a decision. Nobody chose to make your shift end late. The relief hit traffic, the admission came at 18:50, the wall was three units deep. Each one is a reason, and each one is true.

But a shift that ends on a condition instead of a time will run long, on average, forever — not because of any particular night, and not because of anything you did. That's not a complaint about your employer. It's a property of the schedule, and it's the kind of thing that stops feeling personal the moment you can see the shape of it.