Positron emission tomography with (18)F-fluorodeoxyglucose (FDG-PET) has been used to study neurodegenerative disease for the past two decades. This imaging method has conclusively demonstrated that dementing diseases are multifocal, rather than global, brain disorders and has shown that clinically similar dementias may have significantly different patterns of regional glucose metabolism. These observations have helped to focus pathological investigations and provided new insights into disease pathophysiology, yet the diagnostic potential of FDG-PET has not been fully exploited. Now that FDG-PET is becoming widely available to clinicians, it is timely to consider how to evaluate its utility in diagnosing dementing diseases. FDG-PET may be especially valuable in the clinical recognition of frontotemporal dementia (FTD). FTD causes a distinctive pattern of cerebral glucose hypometabolism, but can be difficult to distinguish clinically from more common disorders such as Alzheimer's disease and psychiatric disturbances. Currently available studies primarily evaluate metabolic changes in groups of patients and are inadequate to rely upon when evaluating individuals. Additional carefully designed clinical trials are needed to validate FDG-PET as a diagnostic biomarker.