After a diving accident there are two processes running in parallel: one looking for liability, and one looking to learn. The technical investigation belongs to the second. It rests on preserving the scene and the equipment, recovering the panel and gas data, reconstructing a minute-by-minute timeline, and getting to the causes in the system, not the failure of one person.
Commercial diving lives with a risk that can never be fully removed, and over decades it has built a fairly rigorous way of looking back when something goes wrong. Understanding that process isn't just a matter for supervisors or safety departments: it's part of the trade's culture, and it affects what an ordinary diver does in the minutes after an incident.
This article describes how a technical investigation into a professional diving accident is structured and what's expected of everyone involved. It doesn't go into specific cases or any single country's rules, because the competent authority, the deadlines and the notification obligations change from one jurisdiction to another and should be checked wherever they apply.
First the emergency, then the investigation
Nothing below takes priority over caring for the casualty. The sequence is always the same: the emergency is managed under the planned procedure, the diver is stabilised and evacuated if needed, and only once the situation is under control does reconstructing what happened start to matter.
That said, the investigation starts long before anyone arrives to investigate. It starts the moment someone decides, or fails to decide, to leave things as they are. That's the first point where almost every accident loses information: the team's natural tendency to tidy up, sort out and strip down while waiting for news.
Preserving the scene and the equipment
The rule is simple to state and hard to follow under the emotional weight of the moment: except for rescue or removing an immediate hazard, nothing gets touched.
That includes things that look harmless. Don't shut cylinders or change valve positions. Don't empty or refill cylinders. Don't strip the helmet or the hoses. Don't clean up the work site. Don't switch off the panel. Don't delete or overwrite video or audio recordings. Don't re-coil the umbilical. If something has to be moved for safety, it's photographed first and a note is made of who moved it, when and why.
The equipment involved is isolated, identified and kept in custody until whoever is responsible decides what to do with it. A helmet, a panel or a compressor can end up being analysed in a lab, and their value as evidence depends on being able to show that nobody tampered with them along the way.
The data: what was already on record
A professional diving operation generates far more information than people assume, and a good part of the reconstruction lives there.
- Panel records. Pressures, supply changes, times and the panel operator's notes throughout the dive.
- Gas analysis. The recorded composition of the mixes supplied and traceability of the gas banks used.
- Dive profile. Depths, bottom times, stops and ascents, as recorded by the operation's system.
- Communications. Voice recordings between the diver and the surface, when they exist, which are usually the single most valuable source.
- Video. Footage from the helmet or the work site, plus deck cameras.
- Prior documentation. Permit to work, risk assessment, dive plan, pre-dive meetings, and who signed off on what.
- Registros individuales. Each diver's logbook, with their recent history of exposures, which can matter for understanding the team's accumulated load.
- Maintenance. The history of the equipment involved — inspections, replacements and current certificates.
This list explains why well-run operations insist so much on day-to-day paperwork. On the day you need to reconstruct what happened, the quality of those records decides whether anything can be understood at all.
Interviews: as early as possible, and separately
Human memory degrades quickly and is very susceptible to contamination through conversation. If the team spends the afternoon discussing what happened, by the next day everyone will remember a shared version that looks more like the group's story than what each person actually saw.
That's why interviews are held early and separately, starting by asking for a free, uninterrupted account before moving on to specific questions. Questions focus on observable facts — what you saw, what you heard, what you did, in what order — and avoid asking for opinions on causes or blame, which surface on their own and tend to lead the account astray.
There's an important cultural factor here: people only tell you what they know if they trust that saying it won't cost them their job. Organisations that punish the messenger end up investigating with incomplete information, which condemns them to repeat the same accidents. That's the practical reason, not just the ethical one, for keeping the technical investigation separate from the search for liability.
The timeline: the central tool
The heart of any serious investigation is a timeline. Not a narrative summary, but a sequence of facts timestamped as precisely as possible, each one with its source: this reading comes from the panel, this one from the voice recording, this one from the tender's account, this one from the video.
Building the timeline forces you to separate what's known from what's assumed, and it makes the gaps visible. When two sources contradict each other, the contradiction is noted rather than picking the more convenient version. And when a stretch of time isn't covered by any source, that gap becomes an open question, not a space filled in with whatever seems reasonable.
Once the timeline is closed, the so-called decision points emerge: moments when someone chose between alternatives. That's where the learning is concentrated, and the right question isn't "why did they do that" but "why did that seem like the reasonable choice at the time, given the information they had".
From immediate cause to root cause
Almost every shallow investigation ends at the immediate cause: a hose came loose, the umbilical got tangled, the diver lost supply. That describes the accident, but it doesn't explain it.
Root cause analysis means continuing to ask why, working backwards. If the umbilical got tangled, why didn't umbilical management prevent it? Was that work configuration accounted for in the plan? Was that risk identified in the prior risk assessment? If it was, why didn't the planned control work? Was the scope of the job changed partway through the day without revisiting the plan? Was there schedule pressure? Was the team at full strength and rested?
That chain usually ends up at organisational factors: insufficient planning, procedures that don't reflect how the work is actually done, scope changes that weren't reassessed, uneven training, poor handover communication between shifts, or schedule pressure. These are uncomfortable because they aren't fixed with a reprimand, but by changing how the work is organised.
This way of looking at things didn't originate in diving. The offshore industry absorbed it through disasters, and the most cited of all remains Piper Alpha, whose investigation changed the way safety at sea is understood. The structural lesson was exactly that: serious accidents rarely have a single cause, and there is almost always a chain of earlier decisions that made them possible.
Recommendations that can actually be met
An investigation ends in recommendations, and there's craft in that too. A useful recommendation is specific, assignable to someone, and verifiable: change a procedure at a given point, add a check to the permit to work, modify a piece of equipment's configuration, adjust how a shift is staffed.
Generic recommendations along the lines of "pay more attention" or "raise awareness" are the polite way of changing nothing. If the measure doesn't alter the system, the system will produce the same result again.
Follow-up matters as much as the recommendation itself. Without a date, an owner and later verification, reports end up in a folder and the cycle repeats. Organisations that work well close that loop and audit it.
Lessons learned: collective learning
The final phase is what turns an accident into a shared asset. The commercial diving and offshore industry circulates lessons-learned bulletins: anonymised descriptions of incidents and near misses, with their cause and recommended measures, so other operations can correct course before it's their turn.
They're useful precisely because they don't point a finger at anyone. Strip out the name of the company and the vessel, and the defensive reflex goes with it, leaving what actually matters: a failure mechanism that probably exists in plenty of other operations too. A good deal of what's taken for granted on deck today arrived through this route. Anyone who wants to keep up with that flow of information will find part of it in the industry associations and bodies listed in the directory of companies and organisations, and another part in the everyday conversation of the professional community.
What an ordinary diver can do
Even though others lead the investigation, there are four things within reach of anyone on the team that change the outcome.
The first: don't touch anything beyond what rescue requires. The second: write down what you remember as soon as possible, timed and in the first person, before discussing it with anyone else; a note written in the heat of the moment is worth far more than a polished statement three days later. The third: report what you saw even when it's uncomfortable, including your own mistakes. And the fourth, the most preventive of all: report near misses when they happen, which is the only cheap way the industry has of learning.
In the end, that culture is what separates a professional operation from one that has simply been lucky so far.
Frequently asked questions
Is investigating an accident meant to find someone to blame?
Technical safety investigation and liability investigation are different processes, with different purposes and sometimes different people running them. The first seeks to understand why the system failed so it doesn't happen again; the second establishes legal or contractual liability. Mixing them up makes people stop telling you what they know, which is exactly the opposite of what safety needs.
Who investigates: the company, the client, or the authority?
Usually several at once. The diving company opens its own internal investigation, the client usually opens theirs, and depending on the severity and the country, the labour authority, the maritime authority or an accident investigation body may get involved. Which body has jurisdiction depends entirely on where the incident happened.
Why is it so important not to touch anything?
Because almost all the useful information is fragile. The position of a valve, the contents of a cylinder, the condition of a hose or the configuration of a panel all stop telling the truth the moment someone touches them, even with good intentions. Except for rescue or immediate safety reasons, the scene is preserved.
What are the industry's lessons-learned bulletins?
Anonymous or semi-anonymous publications that describe an incident, its cause and the recommended measures, so other operations can avoid it. They're circulated by industry associations and by companies themselves in their internal bulletins, and they're one of the most useful safety tools precisely because they don't point a finger at anyone.
Is a near miss investigated too?
Yes, and in mature organisations it's investigated with almost the same seriousness. A near miss is an accident that just didn't have the bad luck: the same system failures, without the consequences. Investigating them is the cheapest way to learn.
Published 7 September 2026 · By Arthur More, professional saturation diver
Photo: DVIDSHUB · CC BY via Openverse