Investigations
Usable footage can help reconstruct events and identify suspects.
Evidence from nine countries
International evidence: CCTV and public safety
CCTV can help detect incidents, support responders and provide usable evidence. Results depend on placement, image quality, monitoring, rapid retrieval and the people available to act.
Most examples below concern railways, buses, streets or transport hubs. They inform NSW policy but are not direct tests of cameras at individual roadside bus stops.
View the official petition recordWhat the evidence says
Response
Monitoring is most useful when staff can alert police or transport responders.
Prevention
Targeted cameras paired with patrols and other measures have shown stronger results than passive recording.
Confidence
Visible cameras may reassure some passengers, but privacy and an actual response matter.
Country by country
What has been measured
Each example identifies the evidence type and an important limitation. Operational results show how CCTV was used; they do not, by themselves, prove that cameras prevented crime.
United Kingdom · Observational study
British Transport Police rail crimes
In a study of 251,195 recorded rail crimes, footage was useful in 64.9% of cases where it was available. For robbery, detection was 55.7% when useful footage was available, compared with 8.9% otherwise.
How to read this: This is evidence of investigative value; the comparison does not prove that cameras caused the higher detection rate. Railways differ from roadside bus stops.
Read the source: Ashby, European Journal on Criminal Policy and Research
United States · Randomised trial
Newark: monitoring with directed patrols
A randomised evaluation at 38 camera locations found significant reductions in violent crime and disorder when proactive monitoring was linked to directed police patrols.
How to read this: The trial tests a combined monitoring and policing model, not cameras acting alone.
United States · Independent evaluation
Chicago: variable results
An evaluation reported around 12% less overall crime and 20% less violent crime in the Humboldt Park camera area. Other evaluated locations did not see comparable benefits.
How to read this: The results varied by location and implementation; this is not a universal effect estimate.
Canada · Police operational report
Vancouver: investigation and comparison
Vancouver Police reported footage supporting 55 investigations, including six homicides and 15 stabbings, and approximately 10% less violence at monitored intersections than surrounding intersections.
How to read this: Police operational data cannot establish that cameras caused the difference. Street intersections are indirect evidence for bus stops.
Ireland · Police operational report
Dublin: complex investigation
Gardaí reviewed more than 17,000 hours of footage after the November 2023 disorder. The wider investigation led to 57 arrests and 53 people charged.
How to read this: The outcomes reflect the entire investigation; they cannot all be attributed to CCTV.
Spain · Transport operator report
Barcelona: buses and coordinated policing
Barcelona reported monthly bus offences falling from 605 to 359 between the first quarters of 2025 and 2026, a reduction of more than 40%. The program combined cameras, police activity and action against repeat offenders.
How to read this: The change cannot be assigned to cameras alone, and reporting patterns may have changed.
Germany · Police operational report
Rail stations: usable investigative footage
Following station camera modernisation, German Federal Police reported video-assisted cleared cases rising from about 3,500 in 2019 to more than 8,000 in 2023.
How to read this: Counts rose alongside expanded cameras and review capability; this does not measure crimes prevented. Railway stations are an indirect comparison.
The Netherlands · Independent operational evaluation
Rotterdam: live assessment and de-escalation
A Rotterdam evaluation recorded 8,815 camera-supported de-escalation actions and 3,093 live assessments in 2024.
How to read this: These are operator activity counts, not a measured reduction in crime.
Japan · Observational study
Ichikawa: effects differ by offence
A study of 152 street cameras found vehicle-related crime falling from 1,285 to 887 incidents, about 31%, with generally smaller falls in comparison areas.
How to read this: The study was not randomised and did not find the same pattern consistently for other offences.
China · Quasi-experimental study
City networks and visible offences
A national study of 2014–2019 found faster crime declines where camera networks were denser, particularly for publicly visible offences.
How to read this: China’s policing and surveillance settings differ substantially from NSW; this estimate is not directly transferable.
What NSW can take from this
Choose sites by risk
Assess school travel, reported incidents, isolation, lighting and late-night use. Include regional stops.
Design the whole site
Cover waiting areas and approach routes, improve sightlines and lighting, and check day and night footage quality.
Connect cameras to action
Define monitoring, incident reporting, emergency response, retention and prompt lawful police access before installation.
Protect privacy and measure results
Set access controls, signage and oversight. Publish evaluation of incidents, usable footage, response and passenger experience.
A camera is one part of a safety system. NSW should target locations where risk is greatest and make sure footage is usable and connected to a response.
View the official petition recordThe petition is closed; its official record remains available.
Sources and methodology
This summary uses the linked original studies, government reports and operator publications. The labels distinguish controlled or quasi-experimental research from observational studies and operational reporting. The cases were selected to illustrate potential benefits and practical limits; they are not a pooled estimate of impact at NSW bus stops. Statistics refer to the location, time and outcome stated in each source.
Last reviewed: 23 September 2026. See the NSW evidence and limitations.