Victoria suffered less road deaths in 2003 than in any year since 1949. Both NSW and Queensland also experienced a decline relative to 2002, though 2001 was better still, at least in NSW. In total, about 100 people are alive today who would have been dead if safety had not improved.
As is shown by the contrasting experience of the United States, where road deaths are rising, the long decline in the road toll is primarily due to tighter law enforcement with additional contributions from better roads and cars, and improvements in emergency treatment of people injured in crashes.
Of course, even in the face of decades of evidence, there are plenty who want to quibble.
John Quiggin 
Of course I want to quibble. JQ is quite wrong to tell us that there were less deaths.
He should, of course, have told us there were fewer deaths.
Fixed, thanks (JQ)
The other thing that jumps out at me is ageing.
Most Western countries are ageing faster than the US. Given the preponderance of young men who are the victims and purpetrators of road deaths, it must be a factor. How much of a factor I’ve got no idea.
Just a thought.
US fatality rate PER MILE fell to 1.51 per 100 million miles in 2002 (0.92 per 100 million km). (2003 stats aren’t available yet). That is down from 1.75 in 1992. US rises in absolute death counts are entirely attributable to increased road travel. For comparison Australia’s overall rate is 1.01 per 100 million km, which translates to 1.655 per 100 million miles.
The absolute growth in deaths in the US is almost completely attributable to the “light truck” class, which includes pickups and SUVs. Car numbers have been stable for a decade. (This doesn’t indicate unsafety, as much as it indicates the massive increase in the number of those vehicles on the road. They probably are less safe, or more wrecked, than cars, but that can’t be proved from death counts.)
As I’ve pointed out many times, fatality rates per vehicle-mile or vehicle-kilometre are not an accurate measure of the risks of road deaths since they are, in effect, deaths per driver-kilometre. Risks for passengers and pedestrians are disregarded in this measure.
“Risks for passengers and pedestrians are disregarded in this measure.”
I may have missed this but or forgotten it but I assume you have outlined how to take them into account or referenced a document to that effect.
Can you provide a link or suggest a search that would get me to a description or analysis?