Can you get information about someone's personality from how they look? Obviously people try to do this all the time, but are they getting real information, and how can you tell?
One of my friends - Robin - has, at least in the past, been pretty sure that the answer is "yes", that for instance if someone is a jerk, it shows in their face. I tend to think not. I mean, clearly people use their faces to express themselves, and so there is some connection between expression and personality, but I think the judgments you make when you first meet someone, based on their appearance, only seem to be valid because the judgments themselves color how you see (metaphorically) the person from that point on. That is to say, if I meet a guy and judge by the set of his jaw that he is clever but pompous, I'll tend to keep thinking he's that way whether I have sufficient evidence or not.
When I used to do a lot of chatting with people online, sometimes guys online would ask me, "Are you cute IRL [in real life]?" I would answer honestly, "Not really." They would protest, "But you must be!" I would ask why, and they would say things like, "You sound cute." And I have experienced this myself - it's hard to imagine that an enjoyable online friend is not physically attractive. But there is no relationship between being friendly, or clever, or typing fast, or whatever comes across as "cute" online, and actually being cute to look at.
I realize that's not quite the same question as the face/personality one, but it supports my point (or so it seems to me) that people have confusion over this issue. No doubt it's evolutionarily advantageous to quickly size people up based on their appearances, to the extent possible, but it's also evolutionarily advantageous to appear to be better than you really are (or, in general, to have the most appealing possible appearance regardless of what other characteristics you have). Sounds like a good arms race situation to me.
Then we have this recent study (from UC Santa Barbara), in which women were able to tell, from still face shots of men, which ones liked children most, which ones were the most masculine, etc. The Economist had a pretty good article summarizing the research. I have my doubts about a study that gets so many findings from such a small sample size (something like 30 women and 40 men), but it may be that I'm simply wrong about this issue. It would kind of go along with the way I'm (probably) wrong about sex differences, which I tend not to believe in.