I had some trouble digging up this, so for reference here is a link to reports that list driver (driver only, not other occupants, nor non-occupants) death rates per mile driven.
The major report is called something like The Risk of Dying in One Vehicle Versus Another. It is a pretty large .pdf — if that link doesn’t work, the most current as of this writing was their year 2005, Volume40 Number 3.
It is put out by the Insurance Institute for Highway Safey, a.k.a. IIHS, and auto insurance lobbying group.
Probably the most interesting aspect is the wide variance between models, even within a given weight/size category.
The danger of SUV rollover (to it’s driver) is evident. Giving lie to the myth of SUV safety. The additional danger to other drivers due to extremely heavy SUVs/pickups is not discussed or quantified. I am speaking here of so-called truck-based SUVs whose sales are, as i write this, in freefall due to high gasoline prices. The very existence of GM, and to a lesser extent, Ford is in question due to their belief that gas prices would forever be below $2/gallon. The automakers duped the American (these sorts of vehicles were practically unsaleable anywhere else) public into buy tens of millions of these monstrosities which in addition to obviously using more gas less obviously pollute the air disproportionately due to the grace of pliant politicians. There is currently a plan to bring all SUVs/light trucks to what I would call “pollution parity” (the point at which cars and light truck must meet the same standards) by some ridiculous future date (2015 maybe?) — at which point, barring a return to $2 gas, it will be moot.
The IIHS also has a report on Car-SUV crash compatibility,(Vol 40 No 5) but it didn’t break down certain elements I wanted to see. In particular they note that compatibility between cars and non-truck-based SUVs is better. Yes, quite. But it all boils down to the economics problem of users not paying the full cost of their goods, e.g. excess risk imposed on others is “free”.
That all being said, to take two prominent examples — both of which have well over a million units represented — the Honda Accord and Ford Expedition. The Honda Accord at 56 deaths per million is somewhat better than the 2-wheel drive and somewhat worse that the 4-wheel drive Expedition (rates of 66, and 50 respectively). There is a huge weight disparity between the Accord and the Expedition, ~ 3,500 pounds versus 5,000 pounds respectively. Mid-sized cars are, as the report noted: “safest pound for pound”.
The Berkeley TSC newsletter summarized an October 2004 Journal of Law and Economics article “The ‘Arms Race’ on American Roads: The Effect of SUVs and Pickup Trucks on Traffic Safety. It delves into many size/weight/safety tradeoffs and who pays for what (hint, as usual with cars, externalities are involved); including “The probability of fatalities rises by 45 percent, and serious injuries by 11 percent if a light truck hits a pedestrian or bicyclist rather than a car”