I don't carry a .25 because it really is a pipsqueak cartridge with no stopping power. I have personal knowledge of a case where a husband shot his wife's lover with one, in the face, and the guy lived. The .25 bullet broke his front tooth and lodged in the upper palate, but penetrated no deeper. Needless to say, the guy lived.
Sure you can kill someone with a .25. Even that tiny bullet, if it hits a major artery, or some other vital part of the body, will kill. A .32 or .380 can do the same thing. And it's also true that ALL handguns are underpowered. One of the training videos they show in my department is of a state trooper (I beleive from SC, but I can't remember the state for certain) who got in a shootout at the scene of a traffic stop. He emptied his .357 at the bad guy, and put five out six shots into the torso. The bad guy lived. The bad guy, armed with one of those tiny North American Arms single action .22 revolvers, shot the trooper twice, and killed the trooper. One of the bullets missed the trooper's ballistic vest (it went through the arm opening after it passed through the trooper's arm) and hit the aorta. The trooper bled out internally within two minutes and died at the scene. Two hits with a .22 LR vs. five with a .357 magnum, and the .22 proved more deadly in this case, because even those five torso hits from the .357 failed to hit anything truly vital. That sort of thing happens.
But having said that, a .22, or any really small, underpowered cartridge is still less than optimal. Because while you can find incidents like this, it's still only one incident. When you look at a large number of shootings, you can see a pattern of larger, more powerful calibers tending to be more effective. Put simply, across a large sample of cases, more people will fall to a .45 ACP or .357 magnum than to a .32 or a .25.
This is because a bullet must to two things to incapacitate a human target: it must 1) penetrate deeply enough into the body to hit something vital to the body's ability to continue functioning, and 2) it must do enough damage to that vital something to impair the body's ability to continue functioning, and do it quickly (it doesn't help you if your assailant is mortally wounded but isn't incapacitated before he is able to complete his attack and kill you).
Now bearing all this in mind, and acknowledging that even small, non-expanding bullets can accomplish both these tasks (at least some of the time), and further acknowledging that even larger, heavier, expanding bullets can and do fail to accomplish these two essential tasks some of the time, the fact is that larger, heavier, expanding bullets do the job more consistently, and there are mountains of evidence out there to indicate that this is so.
Now another practical consideration arises: controllability. If bigger is better, why aren't we carrying handguns loaded for the .50BMG? Well obviously, with handguns, there comes a point when the power of the cartridge becomes so great that it impairs the shooter's ability to control the gun, especially for rapid follow up shots if the first misses or fails to incapacitate, and this is why, Dirty Harry notwithstanding, the .44 magnum is not optimal for self defense use.
A couple of centuries' accumulated experience has given us enough evidence that the best balance of power and controllability is achieved with calibers, bullet weights, and velocities within a certain range. Tiny calibers like the .22 LR and .25 ACP fall below that range. Large, powerful cartridges like the .44 magnum and .454 casull exceed it. So while you possibly can (and some people have) successfully defend your life with a tiny little cartridge like the .25 or .22, you'll increase your odds with something a little bigger and more powerful. And when you consider that your one and only life is on the line should you find yourself involved in a gunfight, it really only makes sense to use something more powerful, and stack the odds in your favor as much as you possibly can. The .380 ACP is at the lower end of this optimal range, which is why many call it the minimum acceptable caliber. Perhaps drawing the line here is somewhat arbitrary, and it's certainly true that determining stopping power is at best a very inexact science, which is why debate about it rages to this day. But I think there is enough evidence to support the conclusion that you are best advised to stay within the optimal power/controllability range, in order to increase your odds of prevailing as much as possible.