The following 1995 article (.pdf) from the Journal of Forensic Sciences, and follow on Commentary (also .pdf) from both Dr. Martin Fackler and replies from the article’s authors, might be of interest.
Basically, the authors examined 35 different cranial gunshot wounds, measured them, and tried to draw some correlations between the hole size and bullet diameter. They could tell .22 LR and .25 from .38, but not .22 LR from .25. Fackler then wrote their journal, to tell them that even though they could distinguish between the wounds in their selected cases, their own data showed that you couldn’t make absolute claims about wound diameter and bullet size. Anyway, I thought the interplay was interesting and of value for this particular conversation.
These were wounds made in hard bone, with easily definable holes, FWIW, and there was still potential difficulty in distinguishing between calibers. For one thing, skin is elastic, and can tear unevenly. It is not unknown for the bullet to leave a smaller entrance hole than the projectile’s diameter. Maybe on a pathologist’s table, you could tell this was a .22 vs a .45, and I’d think you could tell the difference between a bullet that had sufficient velocity to cause large permanent wound cavities vs. a bullet that didn’t, but just by looking at the victim on the street? I doubt it. Like some of the results from quantum physics, terminal ballistics can be flat out weird, and counter intuitive.
At page 119 of, Gunshot Wounds: Practical Aspects of Firearms, Ballistics, and Forensic Techniques, Second Edition, by Dr. Vincent J.M. DiMaio, an expert on gunshot wounds, he writes:
The book is fascinating for those with an interest in this subject, and readily searchable via Google Books.
EDIT: Oh, and here’s an old SDMB thread on the same subject.