Greg Mankiw makes excellent points about comparative life-expectancy. He argues that if we standardize the data to correct for factors that determine life expectancy (such lifestyle differences, like obesity, and deaths from non-health injuries, such as deaths by car accidents), we will get a more accurate measure to compare life expectancies across countries.
Update: "Ironman" (great name, BTW) at Political Calculations notes in the comments that he wrote about this very topic back in September and has some very cool interactive tables to help him make this point: when you standardize the data, the US doesn't look all that bad compared to other countries.