In this thread, I'd like to start collecting important papers that revie the state of fMRI in it's various technical and application aspects. o this end, Stephen Strother brought to my attention the following paper by Peter Bandettini:

Bandettini P. Functional MRI today. Int J Psychophysiol. 2007 Feb;63(2):138-45. Epub 2006 Jul 13.

Abstract:
Most brain imaging researchers would agree with the assertion that functional MRI (fMRI) is progressing. Since fMRI began in 1991, the number of people, papers, and abstracts related to fMRI has been increasing; the technology and methodology has shown advances in robustness and sophistication; the physiology of the signal is better understood; and, even though it hasn't yet made significant headway into the clinical setting, applications are widening. Questions that stem from this optimistic and perhaps overly general set of observations include those that ask what the ultimate theoretical and practical limits of fMRI are and how close are we to approaching these limits. In this commentary, I attempt to provide a snapshot of fMRI as it exists at the end of 2005, and to give a clear impression that not only are we progressing by "dotting the i's and crossing the t's" but that fundamental changes in fMRI methodology and processing are being put forth as the field matures.