This study used information in the NSW RTA's Traffic Accident Database System to examine the major patterns of crashes involving heavy trucks in NSW over the period 1996 to 2000 and compared them with crashes involving other road users. Heavy trucks had higher crash rates than seen for all crashes when expressed as rates per registered vehicle, but rates per kilometre travelled, which is probably a better measure of road exposure, did not show great differences between heavy truck crashes and all crashes. Articulated heavy trucks, B-doubles and roadtrains showed higher rates per registered heavy truck for all levels of crash severity compared to rigid trucks and all vehicles but rates per kilometres travelled were only higher for fatal articulated truck crashes. In addition, injury crash rates per registered truck increased over the five years for heavy rigid and heavy articulated trucks but rates per kilometres travelled did not and non-casualty heavy truck crash rates per kilometres travelled decreased overall and for articulated heavy trucks. The patterns of truck crashes reflected the distribution of truck numbers and patterns of road usage across the state and across time. For example, higher proportions of heavy truck crashes occurred in country areas, between midnight and dawn and on higher speed roads. Heavy truck crashes also showed different causal patterns and these differed depending on the severity of the crash. Where the heavy truck was judged to be the vehicle playing the major role in fatal crashes, the most common crash patterns involved an off path on curve movement or were pedestrian-related. In contrast the most common pattern for fatal crashes for other vehicles involved vehicles approaching on the incorrect side of the road from opposite directions. Crashes involving heavy trucks were just as likely to involve fatigue, and slightly more likely to involve speeding compared to crashes involving other vehicles. Alcohol was a component in a very small proportion of crashes compared to other vehicles.