Novel noninvasive functional imaging methods are necessary to predict therapeutic outcome and thereby improve the ability to properly select patients for treatment with both conventional and targeted therapies, to better evaluate therapeutic effectiveness during the early phases of treatment, and to enhance a priori risk assessment for treatment induced toxicity. Functional metabolic imaging typically involves pretreatment baseline magnetic resonance imaging (MRI) and/or positron emission tomographic (PET) scans and performance of subsequent scans during and/or after treatment. Imaging parameter changes are routinely attributed to the intervening therapy and clinical outcomes subsequently correlated with these changes. The physiologic parameter(s) that best correlate with clinical outcome and the relative utility of MRI versus PET are unknown, however. Furthermore, tumor vascular physiology and metabolic parameters are heterogeneous and dynamic processes. Large daily fluctuations often occur in the absence of treatment. The magnitude of this temporal variability is not established for MRI or for PET. Routine and meaningful clinical application of functional imaging requires understanding and quantification of the intrinsic variability of the underlying biologic processes and a demonstration that treatment-induced changes exceed intrinsic temporal variation.