While speaking on a panel at last week's Molecular Medicine Tri-Conference, I mentioned a study from a few years ago that examined the predictive power of a few docking programs. The full reference is Warren GL, et al. "A critical assessment of docking programs and scoring functions." J Med Chem. 2006 Oct 5;49(20):5912-31.
(Note: For those of you interested in a copy of my presentation from the conference, you can download it at https://www.josephcerro.com/docs/20090209_MolMedTriConf.zip (1.8 MB zipped PDF file).