I did a little more research and found this read: Kudos to FullDraw
In his book "Gunshot Wounds" Vincent Di Maio describes various experiments where ammunition was heated in ovens. He says that .22 long rifle cartridges detonate at an average of 275F, .38 Special at 290F and 12 gauge shotgun shells at 387F. The interesting thing about these furnace experiments was that in all instances the cartridge cases ruptured, but the primers did not detonate. In fact the primers were removed from some of the ruptured cases, reloaded into other brass and fired.
When cartridges are placed in a fire he confirms that the most dangerous component of a cartridge is the brass, or fragments thereof that may cause eye injury or penetrate skin, but certainly there is no evidence that a cartridge that is not in a firearm can cause a mortal wound, either by action of the bullet or the brass/primer fragments. It is important to remember however that a chambered cartridge that detonates in a fire is just as dangerous as a cartridge that is fired under normal circumstances in a firearm.
To get a better understanding of the behaviour of free-standing ammunition in a fire, he conducted experiments with a propane torch. A total of 202 cartridges (handgun, centerfire rifle and shotgun cartridges) were used. If the heat was applied directly to the base of a shotgun shell the primer would detonate, the powder would ignite and the shell would rupture. Any pellets that emerged were traveling too slowly to be recorded on a chronograph.
In rifle and handgun cartridges where the flame was applied to the base of the cartridge the primers always detonated but the powder only ignited in half the cases and in those instances the cases did not rupture but the gas was instead vented through the primer hole.
When he heated these same handgun and rifle cartridges at the front, the powder would burn and the cases would usually rupture but with few exceptions the primers did not detonate. The velocity of expelled projectiles ranged from 58 ft/s to 123 ft/s. The only exception was the .270 cartridge where the bullet velocity was 230 ft/s. Primer velocities ranged from 180 ft/s to 830 ft/s.
As a side note he says that a revolver in a fire is especially dangerous because all the cartridges can cook off and be discharged such that there is a danger from projectiles. Only the bullet that came out of the barrel will have rifling marks and the ones that came from non-aligned chambers will have shear marks on them. Obviously if there is a question about the firing of a weapon and whether it was cooked off or fired intentionally they will look for a firing-pin impression on the primer of the suspect cartridge case.
References:
Sciuchetti G.D. Ammunition and fire. American Rifleman 144(3): 36-38, 59-60, March 1996.
Cooking-Off Cartridges. NRA Illustrated Reloading Handbook. Washington, D.C.: The National Rifle Association of America.
And of course Vincent Di Maio's excellent book "Gunshot Wounds - practical aspects of firearms, ballistics and forensic technics".
