One of the nice things about the US, from a research point of view is that, with 50 states, often going in different policy directions, it’s possible to do reasonably good statistical evaluations of policy, taking account of a lot of potentially confounding factors. (It’s also possible to do this badly and dishonestly, as the career of John Lott has shown, but that’s true of any research technique).
Here’s an NYT report of a study, using this approach, which estimates that increases in speed limits in the US led to nearly 1900 extra deaths over 3 years.
Update 27/11 Here’s another NYT story recognising, for the first time AFAIK, that the US trails other countries including Australia in road safety. As commenters on this blog have pointed out, the causes include not only speeding, but also more drink-driving and less effective seat-belt laws. There may also be an effect from the number of SUVs in the fleet, though this is controversial – heavy cars promote safety in some circumstances, and reduce it in others.
On the positive side, not noted here, the US moved earlier to make airbags standard equipment and has much better roads than Australia.
A final point is that the accident measure used here is fatalities per million vehicle miles. This is biased in favor of the US, since it effectively ignores passengers and pedestrians when calculating risks.