The twin studies have provided useful info, but are limited by the small sample size (it's a pretty rare event for twins to be reared separately from birth, and identical twins are an even smaller subset of that group), and the skewed sample (children whose birth parents couldn't or wouldn't raise them are not representative of the general population, and twins have some known and unknown differences from the singleton population). A REALLY weird pattern among twins was recently discovered, and nobody has a clue how to explain it: http://www.medpagetoday.com/OBGYN/2005ASRMMeeting/tb1/1957 If down the road anyone manages to do a study of identical twins gestated in different mothers and/or at different times in the same mother, THAT would be fascinating. And it's technically possible already, but I don't think we'll be seeing this anytime soon in humans -- somebody should try it with animals, though.
There is a lot of info to be gleaned even from the twins who were not separately adopted, but grew up in the same family - and this sample is much bigger. On identical twins one could see what part of IQ variation is provided by strictly individual experiences; on nonidentical twins, after correction for individual variation, one could see what part of the residual variation is due to 50% genetic difference.