It's a popular argument among gun supporters that "guns don't kill people, people do." In the trivially reductive sense, of course, it's true, but that sense is irrelevant to gun laws. In the same sense, smog doesn't kill people -- after all, you have to live in the area and breathe it to suffer injury. The gun supporter might argue that someone has to breathe (at least assuming they don't move out of the area), whereas they don't have to pick up a gun -- but then, none of the kids or teachers at Sandy Hook were the one holding the gun.
That is, much like smog, just existing in the vicinity of a gun puts your life at risk. Or, more pointedly:
Guns do kill people. As sure as any other ambient environmental poison. Guns have legs: toss a hundred guns into a neighborhood, and you can be sure more people will die.
With that in mind, I searched the literature, wondering how good this comparison really was. As it turns out, the necessary calculations are out there. We can make a rough estimate, good enough for conversational purposes, of the comparative effect on mortality of increases or decreases in the density of guns, or the density of smog. After the link, I offer the supporting calculations, and the summatory conversation point for your use.
First, air pollution, which is extensively studied and easy to find data on. According to NOAA, "Poor air quality is responsible in the U.S. for an estimated 50,000 premature deaths each year," and studies suggest that the dose-response relationship is roughly linear at current levels (see here), i.e. a 10% decrease in fin particulates would result in roughly 5000 fewer deaths. This jibes with an EPA projection of about 2000 to 5000 fewer American deaths per year if the ozone standard of 75ppm were dropped by about 10%.
For guns, estimates vary widely. In Hepburn and Hemenway's exhaustive 2004 survey of the literature (it's behind a paywall, but the cite is "Firearm availability and homicide: A review of the literature. Aggression and Violent Behavior 9 (2004) 417-440"), I found several studies that treat handgun ownership as a variable affecting homicide rates with other factors accounted for through multiple regression.
Phillips, Votey and Howell (1976), working near the peak of America's gun violence, found that "a 10% reduction in handgun density would result in a 27% reduction in the homicide offense rate." That's not an error -- guns magnify each other. More guns encourage the carrying and use of further guns in a vicinity, and guns that do kill often kill more than once. On the low end, in a much less violent era, Duggan (2001) found that a 10% increase in the rate of gun ownership was associated with an almost 2% increase in the overall homicide rate." That's all guns, not just handguns, which we would expect to have a stronger relationship.
As a side note, almost every single study in this survey, dozens of pages covering multitudes of studies over decade after decade, found the same conclusion with different measures: more guns kill more people. They increase mortality risk for suicides. They increase attempts at assaults. They cut life expectancy. They positively correlate with homicide rates. On and on and on. Anyone who buys a gun for safety might as well purchase a venemous snake to put in their cabinet.
So what do these estimates give us? Well, in 2009, there were 16,799 homicides in America(CDC page, original stat pdf), so if a 10% reduction in handguns led to a 27% reduction in homicides, it would result in about... 4500 fewer deaths. At the low end, we're looking at 450, and if we concentrated on handguns, a 10% reduction would likely be more like 1000 or more.
Once again: a 10% reduction in air pollutants across America in various categories, at a cost of billions and a not-insignificant retooling of our industrial capacity or of the diesel fleet, would save two to five thousand lives every year. And roughly the same effect could be achieved by just reducing the number of handguns in circulation by the same amount. I'd like to do both. Given a choice, I think I can tell which one would be easier if there were the political will.
Guns have become so numerous in America that it's no longer relevant to think of each one as being in the hands of an owner. Many guns in crime-ridden areas aren't even owned by a single user -- they're "community guns," kept in well-known hiding places and pulled out to be used and returned in short-term criminal acts. Instead of thinking of them as discrete entities, we should act regard them in the same way as we regard air pollution: a lethal miasma generated by countless individual acts, that requires society-wide action to counter the negative externalities of constant badly-incentivized choices.
When you think of guns, think of the gun smog. We dealt with a big chunk of the actual smog problem once, when people realized how much it was hurting them. If we mobilize the same way, we can deal with this one.