How long a work health and safety prosecution takes is invisible in any single announcement. The regulator publishes an outcome, the outcome carries a number, and the years between the incident and that number are not in the headline. We have been counting them anyway, because the delay decides whether an enforcement regime deters anyone or merely records what happened to a business that may no longer exist.
Western Australia is unusually good for this. WorkSafe WA publishes a summary document per conviction, and each one prints both the offence date and the sentencing date in the same table. That makes the lag a measurement rather than an estimate. There are 26 summaries across the 2024, 2025 and 2026 pages. We downloaded and read all of them.
We had a hypothesis, and it is wrong
Reading only the three 2026 convictions, we noticed that the one fatality had reached sentence in 1.97 years while the two non-fatal matters took 3.72 and 3.78. We wrote that three cases cannot carry a rule and put it as a question for the other regulators: whether the seriousness of an outcome pulls a matter forward in the queue.
With the full set, the answer is no.
| Group | Cases | Median lag | Mean |
|---|---|---|---|
| The failure caused a death | 4 | 2.66 yr | 2.54 yr |
| The failure caused serious harm | 7 | 2.63 yr | 2.68 yr |
| Risk only, nobody hurt in the charge | 15 | 2.52 yr | 2.73 yr |
| All convictions | 26 | 2.62 yr | 2.69 yr |
Those three medians sit within seven weeks of each other across a range that runs from 1.97 to 3.78 years. There is no queue-jumping for the worst outcomes, and the 1.97-year fatality that prompted the question turns out to be the fastest case in the entire corpus, not the leading edge of a pattern. It is the sort of thing that looks like a finding at n=3 and evaporates at n=26, which is the whole argument for going back for the rest.
What the wait actually is
The useful number for a duty holder is the spread, not the average. Half of these matters were sentenced within about two years and seven months of the incident. Nineteen of the twenty-six took under three years. None took less than 1.97 years, and none took more than 3.78.
That floor is worth as much as the median. Across 26 consecutive published convictions, not one reached sentence inside two years of the incident. Whatever else is true about WA enforcement, a business involved in a serious incident should expect the legal consequence to arrive somewhere in its third or fourth financial year after the event, and should plan insurance, records and personnel on that basis rather than on the hope of a quick resolution.
This is a different shape from the one we found in New South Wales. There, six convictions announced in a fortnight had not one under three years, and a follow-up a month later found two of four coming in faster. WA's typical case is quicker than the NSW cases we measured. We are not going to turn that into a league table off two samples drawn different ways, and the WA figures are convictions only, which we come back to below.
Where the money lands
Twenty-three of the 26 convictions were of companies, and those fines total $11,048,900. The median company fine is $400,000. The four fatal cases have a median of $886,250, against $350,000 for the rest, so a death does move the penalty sharply even though it does not move the timetable. The largest was $1,750,000. The smallest fine in the set was $18,000.
The other three convictions were of individuals. We have not named them, and this masthead does not, which is the same rule we apply to every court outcome we report: organisations are named, people are not.
The structural point: nearly half are Category 1
Section 31 appears in the charge table of 12 of the 26. That looks alarming if you read Category 1 as the reckless-conduct offence it is in some jurisdictions. In WA it is not. As we set out in that piece, WA's section 31 turns on causation: the failure has to cause death or serious harm, and recklessness is not an element. WA's separate industrial manslaughter crime, section 30A, is where knowledge and disregard live.
The practical consequence is that a serious-injury case in WA routinely draws the top category, which is why nearly half this corpus sits there. Comparing Category 1 counts between states without reading each state's section is a mistake waiting to be made, and the numbers are large enough to make it look meaningful.
What this does not tell you
Three limits, stated because they are the ones that would change the reading.
These are convictions only. The publication page says WorkSafe
does not publish summaries of acquittals
. So this measures how long a successful
prosecution takes, not how long a prosecution takes. Matters that failed, were withdrawn or
are still running are invisible here, and the ones still running are by definition the slow
ones. Every figure above is therefore a floor on the true delay, not an estimate of it.
The year-on-year numbers are not a trend. Medians by sentencing year run 2.30 (2024), 2.75 (2025) and 3.72 (2026), which looks like a system slowing down. We do not believe that reading and we are not publishing it as one: the 2024 page carries only five summaries against 19 for 2025, which suggests the 2024 set is incomplete rather than representative, and 2026 is a part-year with three cases. Ask again in a year.
Twenty-six is small. It is enough to kill a hypothesis built on three, which is what it has done here. It is not enough to support a confident claim about severity having no effect at all, only to say that if such an effect exists it is not large enough to show up across every conviction WA has published in three years.
How we did this
We downloaded all 26 prosecution summary PDFs linked from the WorkSafe WA publication pages for 2024, 2025 and 2026, and read the charge table in each. Offence date and sentencing date are both printed there. Lag is the difference in days divided by 365.25. Where a summary carries several charges, we take the earliest offence date and the latest sentencing date, so each figure is the full span of the matter.
A case counts as having caused a death only where the breach wording says the failure
caused the death of
a worker or person, or the charge is industrial manslaughter. This
matters more than it sounds: the standard Category 2 formula is that the failure
exposes an individual to a risk of death
, and a naive text search for the word death
marks those as fatalities. Our first pass did exactly that and overcounted, which is why the
test is the charge and not the vocabulary.
Two summaries record penalties that are not dollar amounts, a community service order and a global fine across charges, and both are individuals. They are counted in the timing figures and excluded from the money figures. Every dollar figure above is a body corporate fine.
A transcription trap for anyone repeating this: the summaries carry a template line in the footer reading "Prosecution Summary of Conviction" followed by a month and year that is not the document's date. Use the Sentenced Date in the charge table.
We have not contacted WorkSafe WA. This is a reading of published prosecution summaries. Where we have drawn an inference it is labelled as ours, and the hypothesis this piece tests is our own.
Sources
- WorkSafe WA, Prosecution summaries of convictions under the WHS legislation: 2026 (read 28 August 2026): the three 2026 summaries and the note that WorkSafe does not publish summaries of acquittals.
- WorkSafe WA, Prosecution summaries of convictions under the WHS legislation: 2025 (read 28 August 2026): 19 summaries, each downloaded as a PDF and read for its charge table, breach wording, offence date, sentencing date and penalty.
- WorkSafe WA, Prosecution summaries of convictions under the WHS legislation: 2024 (read 28 August 2026): 4 summaries, read the same way.
- Western Australian Parliamentary Counsel's Office, Work Health and Safety Act 2020 (WA), official version as at 31 January 2025 [PCO 00-g0-00] (PDF, 312 pages, read 28 August 2026): section 31 Category 1 and its causation element covering death or serious harm; section 32(1) Category 2 and its risk-exposure wording; and section 30A industrial manslaughter, where knowledge and disregard sit.
See an error? Request a correction and we will check it against the sources above and log the outcome.