I believe the biggest killer of infantry is artillery. I think this is also why a lot of tactical WWII games kind of minimize it - otherwise, the game would just involve a bunch of guys getting blown up by artillery.
I believe 9 mm's loaded with decent hollow points have around an 80% stop rate (stop doesn't necessarily mean kill). .45's with decent hollow points have around a 90% stop rate. With hard ball ammo, those figures drop. I think .45's with hard ball have around a 60% stop rate.
There's a lot of research out there on hand gun ammo effects. Lots of stuff on experiments done on goats, gelatin, etc. as well as real world stats.
Frank Chadwick wrote an article a long time ago called "Lethality in Roleplaying Small Arms Systems", in which he analyzed a number of gun battles between US police and Mexican gangs on the US/Mexico border. They were using revolvers and a few shotguns. Most battles were at close range, in Traveller terms. Some interesting conclusions:
--US police fired 90 shots and hit with 29. This is a very low hit percentage compared with most RPGs, including MGT. Considering that these police would have the equivalent of
at least level-1 in Pistol skill, they hit far less often than their Traveller counterparts. Of 8 shots fired at point blank range (i.e., standing over the target), only 4 hit.
--a "...total of fifteen officers and suspects were wounded ...and they were hit by a total of 32 bullets. Most of the injured men were struck by a single bullet; four were struck by multiple bullets, and of these one was struck by eight bullets. Of these fifteen casualties, 2 were killed almost instantly while the other thirteen survived and recovered. All of those who suffered multiple gunshot wounds recovered. That is, none of the fatalities were caused by a cumulative build-up of trauma, but rather were due to a single, almost instantly fatal, wound."
--In terms of fatality, only 1 of the 11 chest wounds were fatal (!), the single head wound was fatal, and none of the 4 abdomen wounds were fatal.
--Most of the folks who were shot were able to continue functioning. One of the few who was incapacitated with one shot was hit in the wrist and was incapacitated by the pain.
In the famous Gunfight at the OK Corral, about 25 shots were fired, mostly from pistols at 10 feet or less. There were 9 hits (not including the shotgun blast that hit Tom McClaury), implying an overall hit rate of about 40%. However, the real "to hit" percentage might be lower, as Billy Clanton was hit 5 times in rapid succession (he probably wasn't dodging effectively after the first hit) and Tom McClaury was shot once while collapsing from the shotgun blast. So the real hit proportion might have been as low as 20%.
And as Chadwick's study found, most single bullet wounds don't kill the target. Billy Clanton died after taking 5 wounds (though he might have died from fewer wounds). Tom McLaury was mortally wounded by a shotgun blast to the chest. Morgan and Virgil Earp were hit by one bullet each and wounded. Doc Holliday was grazed by one bullet. So, no one at the OK Corral died from a single bullet wound. And Doc Holliday, Morgan Earp and Virgil Earp were able to continue functioning after being shot once each.
So based on this admittedly limited sample, I'd say that a "realistic" RPG combat system should rarely kill or incapacitate a target with one shot. However, there should be some small chance of killing someone outright with a single shot (the numbers above imply that an 8% chance or so of killing the target would be about right). Also, some kind of morale check should be imposed on player characters, since most gun battles (including the ones above) were decided when one side broke and ran. The side that was able to stand and fight prevailed. (This, by the way, was one of Wyatt Earp's most outstanding qualities -- he was apparently extraordinarily cool headed under fire).
Since I run cinematic games, such a combat system wouldn't be terribly attractive to me.