Cognitive impairment in depression may be one of the more practically important aspects of the illness, responsible for much of its morbidity. It also is at the heart of its psychopathology, may contribute to strategies of treatment, and may give us a more easily quantifiable measure of impaired function to correlate with brain activity. Functional magnetic resonance imaging (fMRI) is ideally suited to examine brain function in depression. It has the correct time window to repeatedly sample cognitive task performance; it does not require exposure to radioactive tracers and can therefore be repeated many times; it also can be linked with high resolution structural images acquired in the same imaging session that help identify the regions of activation and support the spatial transformation necessary to compare the scans of different subjects. fMRI has already produced a series of consistent results in depression, identifying increased activity of rostral anterior cingulate and other medial prefrontal structures during effortful tasks and on occasion also increased dorsolateral prefrontal activity, suggesting increased possibly compensatory activity to maintain task performance. Investigating the interplay between limbic (orbitomedial) and 'cognitive' dorsolateral structures clearly has the potential to clarify important illness mechanisms of depressive illness.