Newer cars are generally really safe. Safe bigger cars are likely to be a little safer than safe smaller cars but there are caveats, like needing electronic stability control to reduce the risk of rolling over a bigger car, and needing the car to be designed well to take advantage of its size.
It'd be nice to have an easy way to search the accident databases for how well specific cars have generally done in accidents.
Bike more :-)
Risk of Driving
Car accidents were a leading cause of death in the US in 2011 for my age group (25-44) . That report puts the risk for my age group at 12.4 per 100,000 or 124 micromorts. For comparison, giving birth in the US has 170 micromorts , hang gliding has 8 micromorts per trip , and flipping 20 coins and killing yourself if they all come up heads has about 1 micromort.
That report gives 1,375 micromorts for all causes of death based on that year and age group.
I probably drive less than average (I bike and take the train a lot), but I'd guess that driving is still a leading risk of death for me. I don't have a good intuition for if this absolute level is worth worrying about though.
If I want to worry about it, I can reduce my risk of death by getting a bigger car which will cost more / year in fuel. I should also try to learn which safety features reduce risk by a significant amount so I can prioritize those.
To quantify how much more a heavier car will cost in fuel, I'll compare two: the RAV 4 has a curb weight of 3371 lbs and gets 21 mpg city / 27 highway . The Scion xD weighs 2624 curb and gets 26/32 mpg . If I take the average fuel efficiency for each, assume I'll drive 10,000 miles / year, and assume gas will average $4 / gal, going from 2624 lbs to 3371 lbs will cost $287 / year more in gas .
 http://www.cdc.gov/nchs/data/nvsr/nvsr61/nvsr61_06.pdf (preliminary report) page 30
 http://understandinguncertainty.org/micromorts health tab of the animation
 Spiegelhalter, David (10 February 2009). “230 miles in a car equates to one micromort: The agony and Ecstasy of risk-taking”. The Times (London). Via Wikipedia.
((10000 * 1/24) - (10000 * 1/29)) * 4 = $287
The rest of this page continues with the assumption that driving risk has a high enough ratio of decreased risk per dollar that it's worth investigating whether big cars are safer than small cars.
Big Cars vs Small Cars
“In frontal impacts between cars, the occupants of the heavier car or the one with higher structures tend to fare better than those travelling in lighter, lower cars.”
Here's someone's summary of research from https://skeptics.stackexchange.com/questions/6698/are-big-cars-and-suvs-safer-than-small-cars-or-do-they-just-feel-safer. I take the part about the NHTSA study on 91-99 cars with a grain of salt because technology has changed quite a bit since then.
From the IIHS Status Report, Special Issue: Car Size, Weight and Safety, Vol. 44, No. 4, April 14, 2009 exploring mini vehicles and mid-sized vehicles:
“The heavier car will push the lighter car backward during the impact, which means the velocity change of the heavier car will be much less than that of the lighter car. If the lighter car weighs half as much as the heavier car, the forces on its occupants will be twice as great.”
Also from that report:
“…vehicle size, specifically the distance from the front of a vehicle to its occupant compartment. The longer this is, the lower the forces on the occupants, provided vehicle designers take advantage of the extra length.”
The safety advantage isn't just in collisions with lighter cars, either Single vehicle (solo) crash death rates in in small cars like the Honda Fit are as high in single- as well as multiple-vehicle crashes. The death rate in mini cars (as the report describes the Fit and Toyota Yaris) during 2007 was 35 per million, compared with 11 per million for very large cars - three times as high.
The NHTSA performed a study examining all vehicles up to SUV's, vans, and pickup trucks titled Vehicle Weight, Fatality Risk and Crash Compatibility of Model Year 1991-99 Passenger Cars and Light Trucks. This study showed that:
“Weight reductions in passenger cars, lighter vans, pickup trucks, and sport utility vehicles (SUVs) increased the risk of fatal crash involvement.”
NHTSA did say that the slightly heavier “Mid-Sized SUV's” lost some of their safety advantage because of their increased rollover risk. [Dan's note: they more than “lost some their safety advantage”: older mid-sized SUVs were found more dangerous than large cars] But don't forget, these were older SUV's without stability control. In pure collisions however, they retained an advantage over lighter mid-sized cars because of their higher weight.
Which leads us to rollover risk as it stands today,
SUVs now least likely for rollover crashes From an IIHS report reported:
“Each year, from 2006 through 2009, drivers of newer SUVs suffered an average of 28 deaths per million vehicles, according to the Institute. That's about half the average driver death rate for cars, which was 56.””
About the same IIHS report, in Motor Trend, from 2011:
“The rollover risk in SUVs used to outweigh their size/weight advantage, but that’s no longer the case, thanks to [stability control],” said Anne McCartt, IIHS senior vice president for research, in a statement announcing the report’s findings.”
Also from that report: the IIHS says minivans were safest (25 fatalities per million registered vehicle years), followed by SUVs (28 per million), then pickup trucks (52 per million), with the broad “cars” class rated most deadly (56 deaths per million.)
Stability control was required in 2006 and phased in completely by 2009. When stability control is ubiquitous in passenger cars then this statistic may change. Nevertheless, that the is situation today - SUV's all have stability control and not every car has it.
On rollovers, if you buckle up then the risk is greatly reduced for you, and the rollover risk is largely mitigated. There is also a large variation of risk between specific models.
2% of crashes involved a rollover, but that 2% were responsible for 35% of the fatalities:
“In 2010 alone, more than 7,600 people died in rollover crashes. The majority of them (69%) were not wearing safety belts.”
If you choose to buckle up rollovers are far less of a worry. And don't forget in any case that a rollover less likely to happen now in a SUV than it is to happen in a passenger car!
This paper has reexamined the relationship between light-duty vehicle fuel economy and highway fatalities from 1966 to 2002. Whereas the seminal study by CG concluded that increases in fuel economy led to more traffic-related deaths, this study finds no support for that hypothesis in national time-series data.
This is interesting although doesn't seem to be useful for determining whether a new smaller or bigger car is safer.
From a Consumer Reports article talking about some research (I can't find the research paper itself):
In car vs. SUV head-on crashes, the study found that the odds of death were 7.6 times higher for the car driver than the SUV driver. In crashes where the car had a better front crash-test rating than the SUV did, the car's driver fared a bit better but was still four and a half times more likely to die than the SUV driver.
The FARS is a US database with lots of data on this. I think I can download the detailed data set and get the risk for specific cars rather than the general size categories discussed above. Note to self: general stats like total vehicle miles traveled are on the summary page.