The precise anatomical correlates of ''executive functions'' are still a matter of debate. The assessment of cognitive functions in patients with frontal lobe lesions has led to the term frontal functions'' and ''frontal lobe syndrome'' to characterize patients with disorders of executive functions. The very nature of executive functions implies a process of integration and the coordination of several cognitive processes. A fairly large number of tests are currently used to assess executive functions, but their specificity has not always been clearly established. These tests may at times be affected by lesions of nonfrontal areas or, paradoxically, may be spared by lesions that clearly affect the frontal lobes.' This does not necessarily imply that the tasks are not specific for the evaluation of executive functions, but rather that several nonfrontal structures, neural networks, and subordinate cognitive operations may participate in their performance. The frontal lobes are not a homogeneous anatomical entity, but rather an agglomerate of structures which differ from a cytoarchitectonic viewpoint, reflecting differences in myelogenesis and ontogenesis of the human brain.(2) In addition, from the point of view of anatomy, the frontal lobe represents a ''crossroads'' dense with extrinsic connections that cannot always be taken into account in focal lesions studies. Numerous studies using functional imaging and particularly positron emission tomography (PET) have attempted to define changes that occur at the level of the frontal lobes as a result of pathology, either focal or more widespread. This paper reviews clinical data and previous PET studies and combines them with some new data to show that impairments in executive functions occur in lesions not limited to the frontal lobes and that cortical structures outside the prefrontal areas may play an important role in these functions.