Two-Thirds of Assault Victims Know Their Attacker

Australia recorded 228,544 assault victims in 2025 — more than twenty times the combined number of homicide and robbery victims put together. Assault is, by a wide margin, the largest category of violent crime in the country, and it barely features in the public conversation about crime, which tends to fixate on the rarer, more dramatic offences. The Australian Bureau of Statistics’ Recorded Crime — Victims, Australia release, published 3 September 2026, gives the fullest picture yet of who is actually being victimised, by whom, and where — and several of its findings run directly against the assumptions most people carry into a conversation about neighbourhood safety.

This piece works through five of those findings using the newly released national and state-level data: a three-decade divergence between lethal and sexual violence trends, the true relationship between offender and victim in assault and sexual assault cases, the scale of family and domestic violence hiding inside the headline numbers, a five-year reversal in motor vehicle theft, and a stark and uneven gap in assault rates for Aboriginal and Torres Strait Islander Australians. Read together, they’re less a crime-statistics roundup than a social-cohesion one: most of what shows up in these numbers is happening between people who already know each other, not between strangers — which changes what the data can and can’t tell an investor, buyer or resident about a given place.

A 32-Year Divergence Between Lethal and Sexual Violence

Crime statistics are usually reported as raw counts, but a raw count on its own is a poor way to compare one year with another — a bigger population produces more victims even if the underlying risk to any one person hasn’t changed. The standard fix is the victimisation rate: the number of victims expressed per 100,000 people, which lets a reading from 1993 (population around 17.7 million) be compared honestly with one from 2025 (population around 27.4 million).

Measured this way, Australia’s two most closely watched categories of violent crime have moved in almost exactly opposite directions over the past 32 years. The national homicide victimisation rate has fallen by 60% — from 4.0 per 100,000 in 1993 to 1.6 in 2025 — and the decline has been steady rather than a one-off drop, running through nearly every part of the period. The national sexual assault victimisation rate has more than doubled over the same span, from 69.1 to 144.0 per 100,000, with over half of that increase (a 56% rise) landing since 2015 alone.

By 2025, the homicide rate sits at 40% of its 1993 level; the sexual assault rate sits at more than double it. No single reason explains a gap that size, and the recorded-crime numbers on their own can’t settle the question of how much of the rise is more offending, and how much is more of what was already happening finally reaching police. That distinction matters enough that it’s worth pulling in dedicated research on reporting behaviour, rather than guessing.

What the reporting research actually says

The Australian Bureau of Statistics runs a separate, confidential household survey — the Personal Safety Survey — precisely because police-recorded figures are known to sit well below the true number of incidents. Its most recent sexual violence release, covering 2021–22, found that just 8.3% of women who had experienced sexual assault by a male reported the incident to police; the remaining 92% never went to police at all. That is a strikingly low reporting rate even by the standards of other crime types, which makes the doubling in the recorded rate since 1993 harder, not easier, to read as a simple reflection of “more assaults happening.”

The Australian Institute of Health and Welfare, drawing on the same underlying ABS data, has tracked a second, more direct piece of evidence: the share of reported sexual assaults that relate to an incident from 12 months or more in the past rose from 23% in 2014 to 33% in 2024. A rising share of “old” incidents being reported in any given year mechanically inflates that year’s recorded count without a single new incident having occurred within it — historical abuse coming forward decades later is a well-documented pattern, and it shows up directly in these numbers. The AIHW’s own conclusion is deliberately multi-causal: recorded increases “may be due to changes in reporting behaviour, increased awareness about forms of violence, changes to police practices, and/or an increase in sexual assault incidents” — all four factors are live possibilities, and the data reviewed here can’t isolate one from another.

The honest reading, then, is not “sexual violence has doubled” and not “it’s just more reporting” — it’s that a 32-year, 108% rise in a recorded rate sitting on top of a measured 92% non-reporting rate is a genuinely complicated signal, and any single-sentence explanation of it should be treated with suspicion, including this one.

The Stranger Isn’t the Threat

Ask most people where assault risk comes from and the mental image is a dark street, a pub car park, someone they’ve never met. The data says otherwise. Across the seven states and territories that publish a relationship-of-offender breakdown, 178,911 assault victims were recorded in 2025 with their relationship to the offender captured. Of those, only 23.1% were assaulted by a stranger. Family members accounted for 39.0% of victims and non-family acquaintances a further 26.9% — meaning close to two in three assault victims (65.9%) knew the person who assaulted them. The remaining 11.0% is recorded as “not stated” rather than assigned a category, which ABS itself flags as a genuine residual, not a rounding artefact.

The same relationship-of-offender breakdown for sexual assault specifically shows an even sharper version of the same pattern. Across the same seven states and territories in 2025, 77.4% of sexual assault victims knew their offender — 46.4% through a non-family acquaintance and 31.0% through a family member — while only 13.6% were assaulted by a stranger, roughly one in seven.

Family and domestic violence (FDV) status in this data is assigned by ABS from police recording systems: a victim is classified as FDV-related where the relationship between offender and victim falls within a specified family or domestic category, or where police attached an FDV flag following investigation. Measured this way, 54.5% of all national assault victims in 2025 — 124,518 people — were victims of family and domestic violence. Sexual assault carries a similarly large FDV share: 40.1% of the 39,758 national sexual assault victims in 2025 (15,939 people) were FDV-related.

Put the two findings together and the picture sharpens: household and relationship dynamics, not public-space danger, sit behind the majority of Australia’s two largest violent-crime categories. A locality with a low visible “stranger crime” footprint — the kind of number that shapes a suburb’s public reputation — can still carry substantial underlying assault risk that never shows up in that framing at all, because most of it isn’t happening between strangers in public in the first place.

The Motor Vehicle Theft Cycle Has Turned

Property crime tells a very different kind of story — one of long decline followed by a sharp, sustained reversal. National motor vehicle theft peaked at 139,895 victims in 2001, then fell almost continuously for two decades as vehicle immobiliser technology became standard and car theft grew markedly harder. It reached its lowest point in over 30 years in 2020, at 48,095 victims — a year in which reduced mobility during pandemic restrictions plausibly suppressed opportunistic theft alongside the long technological decline.

Since that 2020 trough, motor vehicle theft has risen in every single year: 49,735 in 2021, 54,985 in 2022, 60,999 in 2023, 65,603 in 2024, and 66,625 in 2025 — a 38.5% increase in five years, with no sign yet of the climb levelling off.

Whether that five-year climb continues, plateaus, or reverses again is exactly the kind of question worth tracking against future releases of this same data — a single year’s uptick is noise, but five consecutive years of increase is a trend that deserves attention.

A Gap That Doesn’t Close Evenly

The release also publishes assault victimisation rates specifically for Aboriginal and Torres Strait Islander Australians in the four states where this breakdown is available: New South Wales, Queensland, South Australia and the Northern Territory (the Northern Territory’s 2025 figures were not published in this release). Comparing each state’s Aboriginal and Torres Strait Islander assault rate against the same state’s general-population rate — the fairest like-for-like comparison the data supports — shows a disparity that is large everywhere and dramatically uneven in its size.

In New South Wales, the Aboriginal and Torres Strait Islander assault rate runs 2.5 times the general-population rate. In Queensland it is 3.7 times. In South Australia it is 5.8 times — more than double the multiple seen in New South Wales, using the exact same measure in the exact same year. A single national “Indigenous Australians face X times the assault rate” figure would flatten a real and substantial difference in how that gap looks state to state, which is itself worth knowing before drawing any conclusion about cause.

A Sharp Divide by Sex in Sexual Assault

Sexual assault is heavily concentrated among female victims. Of the national sexual assault victims recorded by sex and age in 2025, 31,611 were female and 7,546 were male — a ratio of roughly 4.2 female victims for every male victim. That gap is present at every age band the data breaks out, from children through to older adults, and is one of the more stable, long-standing patterns in this entire release; it hasn’t moved meaningfully across the three decades of national data available.

Why It Matters for Reading a Locality

None of this is an argument that crime statistics don’t matter to how a property or a suburb should be understood — it’s an argument for reading them more carefully. A visible, publicly reported crime rate captures exactly the kind of incident it’s built to capture: offences against strangers, in public, that come to police attention quickly. What this release shows is that a large share — arguably the majority, on the assault and sexual assault numbers here — of Australia’s violent crime happens outside that frame entirely: inside households, between people who know each other, and subject to reporting patterns that can take years to fully surface.

For anyone using crime data as one input into how they read a location, the practical takeaway is to treat a low “public” crime footprint as informative about exactly what it measures — visible, stranger-perpetrated incidents — and not as a proxy for the fuller picture this release lays out. The two are related, but they are not the same thing, and the gap between them is bigger than the headline numbers usually let on.

This complements a separate finding on how unevenly that “public” safety picture is distributed even within a single council area — see The Gradient Inside the Gate.

Watching data like this — the underlying releases that describe safety and social cohesion before any of it is scored — is the ongoing work of this desk. None of the figures above are yet incorporated into a certified APN Sentinel™ reading; they’re presented here as source data in their own right. But they’re exactly the kind of evidence this desk exists to track, because the flow-through from a headline crime rate to actual property risk is rarely as direct as the number itself suggests — as the relationship-of-offender and family violence findings above make clear.

Figures in this piece are drawn from the Australian Bureau of Statistics’ Recorded Crime — Victims, Australia release for the 2025 reference year, published 3 September 2026, plus the cited Australian Institute of Health and Welfare and ABS Personal Safety Survey sources on reporting behaviour. All figures are subject to ABS’s own random adjustment to prevent the release of confidential data at small cell sizes; state and national totals for the same offence can differ slightly as a result, which is expected and not a data error.