Depth, Air & Time

Diving Past Your Depth Limit

Every dive computer stops measuring somewhere. Here is what happens at that point, why a computer that keeps guessing is more dangerous than one that goes blank, and how we handle it.

By Mat Mora · Updated 6 August 2026 · ~9 min read

Your dive computer has a maximum depth. Not a maximum it recommends, a maximum at which its sensor physically works. On an Apple Watch Ultra that number is 40 metres, and it comes from the pressure sensor and the EN13319 standard the depth gauge is certified against, not from the app running on top of it. Cross that line and the sensor stops reporting depth. Not a wrong depth: no depth at all.

Most divers will never see this. Forty metres is the recreational limit, deeper than most training goes, and a single cylinder does not give you long down there anyway. But some divers do cross it, usually by a few metres for a few minutes, and what happens in those minutes matters more than their number suggests.

Why the limit exists at all

A depth gauge is a pressure sensor with a certified working range. Past that range the manufacturer will not vouch for the reading, so the sensor reports nothing rather than reporting a number nobody has validated. Apple caps depth reporting at 40 metres for exactly this reason, which is also the recreational diving limit, so the two lines happen to sit in the same place.

This is a hardware and certification boundary, not a software choice. Every Apple Watch dive app hits the same 40 metre wall, whoever wrote it.

What most computers do when you cross it

The common behaviour is a lockout. Oceanic+ raises an alarm past 40 metres and, past roughly 44 metres, stops displaying a depth value at all, showing only that the limit has been exceeded until you ascend back into range. Mares' Apple Watch app is bound by the same hardware limit. Both run Bühlmann decompression models, and both treat out of range as a condition to declare rather than to calculate through.

That is a defensible design. It is honest, it is simple, and it is what certification expects. Its weakness is that it leaves you with nothing in the exact minutes you would most like to have something.

Why a computer that keeps guessing is worse

The tempting fix is to let the computer carry on using the last depth it knew. It is also the most dangerous thing it could do, and the reason is worth understanding even if you never use our app.

Decompression models work on pressure. The deeper you are, the faster your tissues take up nitrogen, and the faster your no-stop time runs down. That relationship is steep. If your computer believes you are at 40 metres while you are actually at 47, it is not slightly wrong. It is crediting you with a shallower, gentler dive than the one you are doing, and every number it shows you afterwards is generous in the same direction.

A blank display tells you something is wrong. A plausible looking number that is wrong in your favour tells you nothing at all. That is the failure mode worth designing against.

How we estimate the depth you reached

Rather than stopping, we estimate, but only in a deliberately pessimistic way. The whole design follows one rule: guessing you were deeper than you really were is safe, because it shortens your no-stop time and brings your stops forward. Guessing you were shallower is dangerous. So we never estimate the most likely depth. We estimate the deepest you plausibly could have been, and run the model on that.

Three numbers we already have are enough to put a firm ceiling on the excursion.

1. How fast you were descending: Your rate of descent at the moment the gauge stopped reading, measured by the watch, not entered by you.
2. How long you were out of range: The gap between the reading stopping and returning, timed to the second.
3. How fast you were ascending: Your rate of ascent as you came back up through the limit.

Those three bound the excursion completely. You went down at a known rate, you came back up at a known rate, and the whole thing fitted inside a known time. The deepest you could possibly have reached is the point where the descent and the ascent exactly fill that time with no bottom time in between. Go any deeper and you could not have got back when you did.

For a short excursion that stays tight, because the situation itself is self limiting. At 40 metres on a single cylinder you are breathing roughly five times your surface rate, so a long deep excursion is not something your gas allows. Cross the limit descending at 6 metres per minute, spend two minutes out of range, and return at 6 metres per minute, and the deepest you could have been is about 46 metres. Stay away for six minutes at the same rates and the bound opens to about 58, and our model gets correspondingly more cautious, which is right, because six minutes away is a genuinely different dive.

Why the bound alone is not enough

Read that table again and you can see where it goes. The bound grows with the time you spend out of range, and it does not stop. Ten minutes away, descending at 9 metres per minute and returning at 20, and the arithmetic says you could have reached 102 metres.

That number is correct and it is nonsense. On air, 102 metres is a partial pressure of oxygen of 2.35 bar against a hard limit of 1.6. You would be breathing about 224 litres a minute, so a 12 litre cylinder at 200 bar is empty in roughly ten minutes with nothing left for the ascent. The dive it describes cannot happen.

Writing "estimated max depth 102 m" into your logbook for a dive that was probably 55 would be obviously wrong to you, and it would take every other number we show you down with it. So the bound is capped by two things.

1. The seabed: Nobody dives below it. We read the seabed depth at your dive's coordinates and cap there. Bathymetry is a coarse global grid rather than a survey of your reef, so the cap carries a tolerance, and a reading shallower than the depth your gauge already measured is discarded rather than applied.
2. Your gas: Below the maximum operating depth of the mix you are breathing, at a ppO2 of 1.6 bar, you are not on a dive any more. On air that is about 66 metres. On EAN32 it is about 40. The richer your mix, the tighter the cap.

A cap does not make the estimate more accurate. It makes it less absurd. That is a real difference, and the app says which one you are looking at.

When the excursion is short enough that your own rates bound it, you get an estimated depth. When a cap decides the number instead, you get an upper bound: a statement about where you could not have been, not about where you were. Those are different sentences and the app writes them differently. The same test decides what reaches the decompression model, because a ceiling on the possible is not something to load tissues against for a whole dive.

What you see on the watch

Your depth returns to a normal reading the moment you are back in range. Your no-stop time only ever moves in the safe direction while the gauge is blind: it can count down, but it cannot go back up. Losing the depth gauge does not mean losing the dive, and everything that does not depend on depth carries on exactly as normal.

What goes in your logbook, and how to correct it

Every dive records whether its depths are real gauge readings, how many times you went past the limit, how long you were beyond it in total, and the estimated depth you reached. Months later your logbook can still explain why one dive's numbers look the way they do, rather than showing a flat line at 40 metres with no hint that anything happened. We record this for every dive mode and every watch we support, so a logbook means the same thing whichever device wrote it.

Your recorded max depth stays exactly what the gauge measured. The estimate sits beside it, labelled, with the reason it says what it says: your own rates, the seabed, your gas, or nothing at all.

And then there is the obvious point. Most people wearing this app are also wearing, or diving next to, a computer that measured the depth properly. That number beats anything we can derive from a gauge that stopped reading, so the dive asks you for it. Tap the estimate, enter the depth you actually reached from your own computer or your buddy's, and the logbook uses yours. It is recorded as yours rather than quietly blended into ours, our estimate is kept underneath, and you can undo it.

This matters for more than tidiness. Your logbook is the record a doctor reads if you ever present with symptoms, and a number nobody can account for is worse than a gap.

This is not permission to go deeper

None of this makes it safe to dive past your computer's limit. An estimate, however carefully bounded, is not a measurement, and a computer that has lost its depth gauge is working with one hand tied behind its back.

Depths beyond recreational limits need the training, the gas and the redundancy that go with them. If you are planning dives near or past 40 metres, plan them properly, dive them with the right kit, and treat your primary computer as the authority on your decompression status. This exists for the diver who did not plan to cross the line and did, not as a reason to cross it.

Nothing here is dive training or medical advice. Dive within your certification and follow your training agency's guidance. Your safety stop is not optional after a deep dive.