I’ve been watching a lot of Mythbusters episodes lately, and they’re always using model heads and bodies made out of ballistics gel, under the premise that ballistics gel “reacts similarly to a human body under these circumstances.” They do sometimes embed a pig spine or something in the ballistics gel model for verisimilitude (frex, they did this in the "Will a truck tread decapitate you? myth). But really, overall is this stuff that close to human flesh/skin/bones/etc.?
Properly calibrated 10% ballistic gelatin is a very close analogue in terms of penetration to human muscle and solid organ tissue. The general rule of thumb is that it compares favorably in penetration value to real world application of a non-protected target at about 105%-110%; that is, 11" of penetration into ballistic gelatin is about 10-10.5" in a real body. Since human skin is very thin and of essentially the same consistency as muscle tissue, skin doesn’t effect results, but leather, wallboard, neoprene, and other thick, semi-conformal coverings can plug the hole of hollowpoint bullets, causing them not to expand reliably or at all.
A bullet that passes through a hard substance (like glass) before hitting the body, or one that strikes solid bone will deform or fragment and its travel in the body will be erratic, so it’s hard to characterize the general behavior of bullets in this case. Nominal expansion doesn’t offer much in the way of increased damage–a slightly larger permanent wound channel, but the real world stopping performance between expanded and unexpanding bullets of similar penetration shows only a very marginal improvement. The major benefit to expansion is that the bullet is more likely to stop within the body, or at least deliver the bulk of its momentum, and thus not pose as much of a threat to bystanders. Fragmentation in terminal ballistics, on the other hand, offers a marked improvement in stopping potential. This assumes, however, sufficient penetration; the problem with prefragmented rounds like Glaser Safety Slugs is that their penetration is rarely deep enough to strike at critical nervous system points or interfere with major organs, instead making bloody but superficial wounds. (They also don’t cycle well in many semi-automatics.)
Ballistic gelatin, BTW, is basically Jello without the sugar. It’s not hard to make–provided you can clear out your fridge–but because its such a pain in the ass to handle and set up, most amateur enthusiasts I used to know preferred telephone books soaked in soapy water. The corollation of this method is somewhat questionable (everybody seems to have a differing rule of thumb on what penetration in soapy phone books equals in the real world), but it did seem reasonable consistant and reproducible.
Stranger
I have not handled ballistics gel, but I’ve had my hands in a lot of people. Judging by the way the gel appears to behave on Mythbusters, I would guess that while it may mimic muscle well, it is far too firm to model softer tissue like fat. It doesn’t seem to sag or drape like normal flesh does. I also suspect that it does not have the resistance to the propagation of a tear that living tissue does. You can put a few hooks in your skin and suspend yourself, but I bet the hooks would tear out of a gel dummy.
So, depending on what behavior of human tissue you plan to mimic, I would guess that it could be a fair mimic or a poor one. I suppose that if they were really going to be anal about it they could embed bones in a relatively thin layer of ballistics gel and then zip the thing into a person-sized worn-out spandex bodystocking and inject some sort of a fluffy chopped gel mixture under the spandex to simulate fat.
In fact, they do sometimes go that extra mile to increase the simulation of a human body. And other times they abandon the gel and go with a pig carcass.
Thanks for the excellent information, everybody. 
It confirms my suspicion, which is that ballistics gel is an okay but not perfect substitute for a human body in a lot of these experiments.
One thing ballistics gel does perfectly, when used as part of an explosives test, is identify the frag pattern. This is at least as an important part of the experiment as knowing the penetration capabilities.