Positron Emission Tomography (PET) has allowed researchers to examine in vivo brain-behavior relationships. Correlations of metabolic increases measured by PET with simple sensory tasks involving hearing, vision, and tactile/motor responses have generally produced metabolic changes in agreement with known neuronal pathways. However, complex neuropsychological tests such as Raven's Matrices and Verbal Fluency have resulted in cortical activation of unexpected areas as well as some negative correlations between test performance and metabolism. These observations provide the first demonstration of complex relationships between neuropsychological functioning and activation of the normal brain. Since PET technology offers a means of computing the simultaneous proportional contribution of multiple brain regions, the issue of "localization" of neuropsychological test performance might appropriately be considered for redefinition in terms of a "systems" approach. In spite of the advances brought about by PET, methodological problems still remain for researchers. Future studies will need to take into account the subtle regional effects of neuropsychological paradigms.