Abstract:  This talk provides new evidence on a fundamental issue in the U.S. debate over how to regulate guns.  The issue is illustrated by a thought experiment: If violent people no longer had access to guns, how much would the murder rate fall?  (Currently, 70% of murders are committed with guns.)   Opponents of gun control, including “experts,” assert that there would be little change, since killers would simply substitute other weapons with the same result.  The new evidence supports the contrary view that the type of weapon has a large independent causal effect on the outcome of an attack.   We demonstrate using a 5-year sample of shootings in Boston that the power of the gun (as indicated by its caliber) is highly correlated with the probability of death.  That this correlation is causal is suggested by the fact that caliber of gun is uncorrelated with all observable characteristics of the attack.