I chose Davis over UCSD and Irvine, and thought I'd share why for future applicants.
Interests: Cosmology Experiment/Simulation (CMB, weak lensing, etc)
Caveat: I have a significant other whose opinion weighed heavily into the decision. I'm pretty sure that I'd have picked the same regardless, but one never knows such things for certain.
I first narrowed the choice to UC Davis, UC Irvine, and UCSD. California schools generally attract solid Astrophysics departments (because of Keck), plus I was getting tired of the east coast. After visiting, here was my impression of each school:
Very much in the OC suburbia. Strip malls abound. It's not like you can't have fun there, it's just that there's nothing interesting about it. Definitely something I could have gotten over for the right school, which for a lot of people Irvine could be. For my interests they had on paper an excellent program, with Elizabeth Barton, James Bullock, Asantha Cooray, and Manoj Kaplinghat all relatively young, promising, and productive researchers. However, Barton is leaving, and there doesn't seem to be a lot of depth to their observation faculty. Also, there just didn't seem to be a collaborative environment within the subfield. Cooray was rarely around (spending loads of time at Cal Tech) and because of OC traffic people kept weird hours. I think this would conflict with the daily sharing of ideas, etc. Personally, I also didn't really hit it off with any one group; the only research I was excited about was Cooray's, and I'm afraid he might follow Barton out the door to Cal Tech at first opportunity (he has a little bit too big of a profile for a place like Irvine). Note: UC Irvine is expensive, but the department's policy is a 100% RA over the summer, so that adds about $6500 to the stipend for experimentalists (theorists, like theorists everywhere, don't have money).
UC San Diego
This department really boiled down to Brian Keating. It's a top 15 school, so there's ton of stuff going on, but in Astro, it seems to revolve around him. He's got by far the biggest group, the most money, and quite possibly the biggest profile. Which was fine with me, since CMB (his field) was pretty much exactly what I wanted to do. In all, a decision for grad school based on maximizing my chance for an R1 faculty job would have led me to UCSD. But other things counted heavily against it. Firstly, SD isn't a great place to be a grad student. the ~23k/year base stipend is poverty there (UCSD is extremely expensive). To get anywhere reasonable priced, you need a car. Somewhat mitigated by grad school housing, but you can only live there for 2 years. I also got the impression that the group was more focused on instrumentation, which I abhor (I was promised analysis work, and a spot in the group, without these I wouldn't have considered the school at all). This was a really tough one to turn down, which is why my decision took so long.
I was really impressed by my visit. The 5th floor is just cosmology, and the relative isolation is a really good thing: there was tons of informal collaboration just happening in the hallway. All the grad students in cosmology are in 3 rooms next to each other, and Davis has done a good job of bringing together faculty with similar enough research to provide really strong collaborative effect. Everyone was also really friendly; when I visited professors would drag me into their office to discuss research anytime they noticed me in the hallway. My probable research advisor, Lloyd Knox, is situated on the experiment-theory continuum just where I wanted; he's very close to experiment (working inside the experimental group for SPT and personally being granted time on Herschel) while still less concerned about retrieving the signal than interpreting it. He also just got a ton of money, so that's always good. Also, the city of Davis made me happy; you can bike literally everywhere in town, and it's only a short drive to the Bay, Napa, and Lake Tahoe. Since I'm signing on for ~10% of my life there, it's important that I enjoy my time.