fMRI has unique potential in the study of psychiatric patients, particularly in characterizing individual variations and changes over time. We have performed four studies of patients with schizophrenia, using three different fMRI acquisition protocols: (1) 3-D echo-shifted FLASH, a multishot volumetric approach; (2) 3-D PRESTO, a hybid of multishot and echo-planar imaging (EPI) methods that also acquires true volumetric data; and (3) a whole-brain isotropic, multislice EPI technique. Patients were studied during sensorimotor activation and during a novel "N back" working memory paradigm. In general, patients show normal sensorimotor activation responses, although motor cortical activation tends to be less completely lateralized. Prefrontal activation during working memory tends to be reduced in patients with schizophrenia even when performance is normal. A major potential confound in studying this patient population with fMRI is the effect of motion. We propose several methodological standards to address this problem, including comparisons of motion corrections parameters, voxel variances, and the use of an "internal activation standard."