Peter D. Kramer (of "Listening to Prozac" fame) has a relatively new book out called "Ordinarily Well: the Case for Antidepressants". I felt like if you do the math yourself for the studies he cites antidepressants didn't come out that well, but he obviously feels different about them. It's worth reading anyway.
For me the only antidepressant that worked was Wellbutrin.